The stem form brahma found here and in many other compounds in the KhV is ambiguous: it can denote the deity Brahmā or the ultimate reality brahman. In tantric texts and early works of hahayoga it usually refers to the deity (as in the system of the three granthis, brahma°, vi and rudra°, at e.g.HP 4.70–76; cf.KhV 3.3b where dhāma svāyabhuva is used as a synonym of brahmadhāma). The inherent ambiguity allows later authors to interpret such compounds in a Vedantic light: e.g. HPJ 3.106 where brahmasthānam is glossed by Brahmānanda with brahmāvirbhāvajanakam sthānam, “the place that reveals brahman”. I have chosen to translate brahma° as Brahmā.

241 “the region above the nape of the neck” (cūlitalam): as far as I am aware, cūlitala occurs only in the MaSa (9.24a), the KhV and derivative texts. From the evidence of 2.49–59 it appears to mean the region above the nape of the neck, on the same level as the forehead and temples. This meaning fits well with the context here. Ballāla (f. 21r67) agrees, taking cūli as a variant form of ā, “the crown of the head”, and tala as meaning “the area below”: tiryak cūlitala śikhādhobhāga yāti cūli śikhā / śikhā cūā ity amara / alayor abheda. See also 2.18 and note 285.

242 See note 239.

243 “not all at once” (yugapan na hi): Ballāla (f. 22r3) glosses yugapat with ekasamayāvachedena, “cutting [the frenum] all at once”. One of my informants, Govind Dās Jī Mahātyāgī, did cut his frenum all at once. He told me that the cut bled a great deal but that otherwise he had no problems. See also 1.46–48 and the notes thereon. Ballāla adds (f. 22r23): yady api abhyāsakāle kadā cid asvāsthya tadā taddine heyo ’bhyāso ’nyadine susthatāyā kartavyo na jhaiti, “if ill health should ever arise during the practice then it should be abandoned for that day and taken up on another day when good health has returned, not straight away”.

244 brahmabila is synonymous with brahmarandhra in its first sense (see note 240).

245 “the bolt [of the doorway] of Brahmā”(brahmārgalam): Ballāla (f. 22r9) equates the brahmārgala with the brahmadvāra: brahmārgala brahmamārgapratibadhaka rājadatordhvadvāra, “the brahmārgala is the door above the uvula which blocks the pathway of Brahmā” (on the rājadanta see note 258). In the text, however, the two seem to be distinguished. The brahmārgala, “the bolt”, is to be rubbed away for three years, after which time the tongue enters the brahmadvāra, “the door”. 2.1a and 3.44 mention the brahmārgaladvāra, “the bolted door of Brahmā”. In descriptions of the goddess Kualinī she is often said to be asleep blocking the brahmadvāra at the base of Suumā and this is its usual location (HP 3.5, GŚN 47, YCU 37,. CN 3 and 50). In the KhV the brahmadvāra is at the other end of Suumā, at the opening at the base of the palate.

1.55c–57b is puzzling. After a total of seven years the yogin is instructed to start rubbing at the brahmārgala so that after a further three years the tongue might enter the brahmadvāra. This is the first time in the section on the physical practice that the yogin is told to try to turn his tongue back. As I have noted at 1.46a it is possible to insert the tongue into the cavity above the soft palate without any preparation. So what is the internal destination for a tongue that externally can reach the crown of the head? The cavity above the soft palate is surrounded by bone so it would seem that however much rubbing the yogin may do there is nowhere else for the tongue to go. And why should the yogin wait so long before turning back his tongue? Are the verses that describe the extreme extension of the tongue so much arthavāda, designed to put off prospective khecarīsiddhas? Or did some yogins actually lengthen their tongues this much in displays of ascetic self-mortification? None of my informants had particularly long tongues yet most claimed that they had perfected the practice. I have heard of one yogin, Sampat Nāth of Ajmer, Rajasthan, whose tongue could reach his bhrūmadhya (personal communication from Robin BROWN, 1996). No other text (except the MKS whose description derives from that of the Khecarīvidyā) claims that such extreme lengthening of the tongue is necessary to practise khecarīmudrā. HP 3.32 states chedanacālanadohai kalā kramea vardhayet tāvat / sā yāvad bhrūmadhya spśati tadā khecarīsiddhi // “By means of cutting, manipulation and milking [the yogin] should gradually lengthen the tongue until it touches the centre of the eyebrows. Then [there is] khecarīsiddhi.” Cf.KhV 1.73ab, where the siddhis brought about by means of the practice are said to arise between the eyebrows. The two other texts that deem the cutting of the frenum necessary for the perfection of khecarīmudrā, the Haharatnāvalī (2.130–142) and the Gheraasahitā (3.21–22), also state that the tongue need only be lengthened enough for it to reach the region between the eyebrows. There is one ancillary benefit of lengthening the tongue: it can be used internally to control which nostril the yogin is breathing through, thus eliminating the need to use the hands during prāāyāma. This was reported to me by several of the yogins I met during my fieldwork and is described by BERNARD (1982:68).

246 It is hard for the gods to pierce “because they are intent on pleasure” (bhogāsaktatvāt): Ballāla f. 22r9.

247 “the door of Brahmā” (brahmadvāra): in the Khecarīvidyā, brahmarandhra, brahmabila and brahmadvāra seem to be synonymous (see notes 240 and 244). KAVIRĀJ (1987:51) reports that in the Vairāapurāa the brahmadvāracakra is above the fore-head but below the brahmarandhra in the cranium.

248 “churning” (mathana): where the word mathana occurs in the text (1.57–63, 2.101–104), witnesses µSM occasionally, but not consistently, read mathana. This reflects the two forms that the root can take: art and art (WHITNEY 1988:117).

249 Ballāla (f. 23r34) explains that the thread is passed through a small hole in the probe, like that in a needle: tena [sūcyām iva] śalākānuchidre protenety artha /.

250 It is not clear to me how this practice is to be carried out. Ballāla adds little to what is found in the text, thereby indicating that he too is unfamiliar with the practice. In his commentary on 2.101 (f. 58r4) he explains the purpose of practising mathana as sarvamalaśodhanārtham, “to cleanse away all impurity”. This is clearly not the main aim of the practice since 1.64 says that mathana brings about sasiddhi and identification of body and self with the universe. In the Khecarīvidyā, it appears that after the probe is inserted into the nasal cavity it is to be moved about by the tongue, which has entered the cavity via the palate. The word mathana usually refers either to the rubbing of wood to produce fire, particularly in a sacrificial context, or to the churning of milk to produce butter. It is used in this second sense in the archaic myth of the churning of the ocean of milk by the devas and asuras (MBh 1.17–19; cf. SYM 21.7, 22.36; see also GONDA 1965:61). Both senses of the word seem applicable here. Firstly, at KhV 2.72–75 the yogin is instructed to churn the circle of fire (mathitvā maala vahne) at the base of the tongue (jihvāmūle) and thereby melt the orb of the moon into amta (cf.HP 3.48). MONIER-WILLIAMS [1988:s.v.] reports that śalākā can mean “a match or thin piece of wood (used for ignition by friction)”. Secondly, when the ocean of milk was churned, amta was among the fourteen items that were produced. No yogin that I have met practises mathana as described here but Dr. hākur of Mumbai did describe how during his practice of prāāyāma and khecarīmudrā his tongue would involuntarily start to “bang away like a drill going into a hole”.

The Jogpradīpakā has the most coherent description of the hahayogic mathana and describes two varieties. In the first (v. 642) the yogin is to rub the śivaliga, which is also the agnisthāna, at the root of the palate with his thumb three times a day. In the second (vv. 643–653), the yogin is to use a metal peg (kīla dhātamaya) to churn, purify and produce amī (=amta) at four places: ambikā, the frenum, lambikā, the tongue, tālu, the palate and ghaikā, the uvula. These four places are said to be the teats of Kāmadhenu (v. 651). Cf. the Vairāapurāa, which locates an amtacakra in the upper part of the forehead from which “nectar is constantly flowing. This place is described as the abode of the Gāyatrī named Kāmadhenu (the ‘wish-giving cow’) figured like a milch-cow with four teats, viz. Ambikā, Lambikā, Ghaikā and Tālikā” (KAVIRĀJ 1987:50). The Amaraughaśāsana, describing practices akin to the hahayogic khecarīmudrā, mentions kalāpamathana, “tongue churning”, (p.2 1.10, 1.13). It says that the practice brings about īmukhojjmbhaam, “opening of the mouth of the [śakhinī] nāī”, but does not go into detail. MaSa 27.7 describes massaging the body with a preparation which has among its ingredients amta that is mathanaja, “produced by churning”, and brahmarandhravinirgata, “issued forth from the aperture of Brahmā” (see note 328). At SSP 2.8 the nirvāacakra, which is situated at the brahmarandhra, is described as sūcikāgravedhya, “to be pierced with the tip of a needle”; this, however, sounds more like cranial trepanning than the practice of mathana described in the Khecarīvidyā. The SSP’s reading °vedhya is uncertain. A variant °lekha is found in one witness. The passage also appears at Saubhāgyalakmyupaniad 3.8 where we find sūcikāghetara and a paraphrase at ŚP 4359 reads sūcikāgrābham. Abhinavagupta ( 5.22–24) describes an internal mathana in which apāna and prāa are churned to force the breath upwards into Suumā and ignite udāna. In one of the earliest textual references to Kualinī, the c. eighth-century CE Tantrasadbhāva 1.214c–228d (NGMPP reel No. A188/22; the passage is quoted by Kemarāja ad Śivasūtravimarśinī 2.3 and Jayaratha ad TĀ 3.67) describes a technique whereby Kualī is awakened through the churning of bindu. KMT 12.60–67 describes a mathana at the maipūra centre, in which the churning of a liga in a yoni results in ajñānamalanāśana, the kindling of jñānāgni, a bliss like that generated in sexual intercourse and, finally, amta, with which the yogin is to visualise his body being flooded. On the Khecarīvidyā’s corporealisation of these subtle practices, see page 27.

251 In the texts of hahayoga, the jīva is the vital principle, entering the fetus at the moment of conception (SSP 1.68) and leaving with the body's final exhalation (YCU 90). It moves about the body, propelled by the breath (GŚN 38–39), unless restrained by means of prāāyāma (GŚN 40–41). Ballāla (f. 23r10) glosses jīva with prāa which seems to be an oversimplification: GŚN 37 describes the ten vāyus as flowing through the īs while “having the form of the jīva(jīvarūpina). Cf. ŚP 4317. See also KhV 3.34-46, VS 5.4–7, ŚS 2.39–53, ŚP 4503–4504, TŚBM 60-62b, KJN 6.1–14 and YBD 3.253–293 for descriptions of the workings of the jīva.

252 Ballāla (f. 23v8) says that churning is not meant to be done constantly “because it is very difficult” (kahinataratvāt).

253 BERNARD (1982:68) reports that he kept his tongue in the cavity above the soft palate at all times, removing it only “to speak, eat, or engage in some other activity that made its position inconvenient”. Cf. KJN 6.25c–26d, GhS 3.7. Lāl Jī Bhāī told me that khecarīmudrā should be practised for two to three hours a day. The “pathway” is the pathway mentioned at 1.65a.

254 Ballāla (f. 24r38) understands this to mean twelve years from the time of first cutting the frenum, thus equalling the time needed to achieve siddhi mentioned at 1.70. He reckons the various stages of the practice up to the perfection of mathana to total eight and a half years (in my edition they total ten and a half years), thus leaving three and a half to wait for sasiddhi. The Jogpradīpakā (v. 613) also teaches twelve years as the amount of time needed to master khecarīmudrā.

255 Cf. AY 1.95ab: brahmāa sakala paśyet karastham iva mauktikam, “he sees the entire universe like a pearl in [his] hand”; see also KJN 14.62–65.

256 “the great pathway” (mahāmārga): this refers to the top of Suumā (cf. HP 3.4 where mahāpatha is given as a synonym of Suumā).

257 “in the skull” (brahmāe): in the KhV brahmāa means “skull” rather than the more usual “macrocosm”; see 2.36, 2.42, 2.67c–68b, 3.16–17d; cf. SCN 53d; AM 8.1, p.109; ŚS 1.51, 1.95, 1.97, 1.99, 2.5, 2.37, 3.9, 5.191 (in which the physical body is called brahmāa); GBS 217; Kathāsaritsāgara 2.15; SHEA & TROYER (1843: 132) “the seventh region is that of the head, which is called by the Hindus brahmāa”. Ballāla (f. 24r9), however, takes it to mean “macrocosm”. Later Sanskrit and hahayogic works have a system of 21 brahmāas in (and above) the head. See GBP 19.0 and the Vairāapurāa (KAVIRĀJ 1987:52). At 4.133cd brahmāa (understood to mean the universe by Jayaratha ad loc.) is said to arise from the sahasrāra cakra at the top of the head.

258 “in the region above the uvula” (rājadantordhvamaale): the Royal Tooth (rājadanta) is the uvula. SSP 2.6 locates it at the tālucakra, equating it with the ghaikāliga, the mūlarandhra and the “tenth door” (daśamadvāra), which is the opening of the śakhinī nāī (on which see note 131). Ballāla interprets rājadanta in two ways: firstly (f. 24v13), it is the microcosmic equivalent of the macrocosmic Prayāgarāja; he thus seems to be putting it in the same place as trikūa (see note 259) when the text clearly states that it is below trikūa. Perhaps trikūa can be thought of as a peak above the confluence. Secondly, “some say” (ke cit), in the body the rājadanta is the uvula (f. 24v11–f. 25r1): he describes it as a hanging piece of flesh (māsalolaka) in the area above the root of the tongue (jihvāmūlordhvabhāge) like the clapper of a bell (ghaālolakavat)—cf. note 236. See also 2.29cd, 3.16c–17b and pp.10–11, GŚN 147, HP 1.46, 3.21, ŚS 3.84, KJN 6.23, KMT 9.82, 23.167. Taittirīyopaniad 1.6.1 calls the uvula indrayoni, “the source of Indra”.

259 “the Three-peaked Mountain” (trikūam): this passage and 3.16–17 locate trikūa between the eyebrows; a variant reading at SSP 6.81 locates it at the brahmarandhra; see also KhV 2.81c, SSP 3.5, YV 20 ( BVU 73), AM 85.2, GBP 11.2. MBh Sabhaparvan 2.39.11cd implies that trikūa is in the forehead: lalāasthā trikūasthā ga tripathagām iva; MBh Bhīmaparvan supplement 6.3.88 locates it at the base of the palate: tālumūle ca lampāyā trikūa tripathāntaram. Ballāla (f. 24v14) continues the theme of micro°/macrocosmic equivalence and takes mahāmārga to mean the rivers Gagā, Yamunā and Sarasvatī. Thus trikūa, where the Iā, Pigalā and Sarasvatī īs meet, is the bodily equivalent of the confluence of the three rivers, the triveīsagama, located at Prayāgarāja (the modern-day Allahabad). He explains trikūa as meaning trayāā mārgāāa, “the peak of the three ways” (f. 25r57). At f. 25r34 he cites ŚS 5.132, where the conjunction of the three īs is equated with the confluence in Vārāasī of the Gagā with the Varaa and Asi rivers.

At f. 25r78 he cites a list of esoteric centres which he ascribes to the Kapilagītā of the Pādmapurāa: trikūa śrīhahasthāna golhāa auapīhaka // pūrādri (corrected in the margin from puyādri) bhrāmarīguphā brahmarandhram anukramād iti. In the margin of f. 25r is a quote attributed to Goraka in which trikūa is located at the mouth: asyārtho gorakea darśito yathā // mukha trikūam ākhyāta pthvītatvam ācāraliga gveda[] brahmadaivatam īśvara pītavara jāgrad- (em. ISAACSON & GOODALL; jāgd S) avasthā sthūladeha[] iti / śrīhahasthāna rasanā’pastatva guruliga yajurveda svapnāvasthā viur deva[] śvetavara tatvam iti / golhāa tu nayanasthāna tejastatva śivaliga sāmaveda suuptāvasthā rudradeva raktavara trimātrādehasabhavam iti / pūrapīha ca nāsikauhapīhasajñaka / pādatatva ir vāyur jagamaliga (em. ISAACSON & GOODALL; jagama liga° S) daivata / atharvaveda (em.; atharveda S) turīyā ca okāra nīlavarakam iti / puyādrir (em.; puyādir S) merur ity artha / bhrāmarīguphā śrotrasthāna ākāśa i prāsādaliga sūkmavedaka unmanī śivaliga kavaram iti / brahmaradhre sahasrāre daśamadvāre sarvatatva tanmātrāśabdasparśādipacaka // caitanya sūkmadeha ca parabrahmātmaka mahad iti /. I have been unable to find this passage in any other text. It is the most detailed description of these esoteric centres and their locations that I have come across. This system is usually found only in texts from the Marāhī-speaking region: a similar, but less detailed, passage is found at AM 42; see also AM 55.2, 63.2; SSP 2.27, 6.81–82; YV 20–21; VD 10. Some lists of śāktapīhas in SIRCAR (1998:s.v.) include Trikūa, Śrīhaha and Pūragiri, while the goddess Bhrāmarī is associated with a ha called Janasthāna whose microcosmic location is the chin. (As Alexis SANDERSON has suggested to me, Janasthāna may well be wrongly written for Jālasthāna (=Jālandhara).) The brahmarandhra’s macrocosmic location is Hig Lāj in Baluchistan. Of the bodily centres listed in the BKhP’s citations quoted above, only golhāa and auapīha are not listed by SIRCAR (1998) as geographical locations. SANDERSON has suggested that they are variant spellings of kollāa (=Kolhāpur?) and auapīha (=Oiyāna). A bodily centre called gollāamaapa is mentioned at SSP 2.27 with variants kolhāa and kollāa. The tentative identification of Kollāa with Kolhāpur is supported by a description of female Maharashtrian entertainers called Kolhāanīs by SONTHEIMER (1989:236).

A work entitled Trikūārahasya in the MS collection of the Bodleian Library (Chandra Shum Shere e.83(4)) describes tantric ritual of the Śrīvidyā tradition.

260 Jogpradīpakā 651 compares the uvula (ghaikā) to a chickpea sprout.

261 See Kathāsaritsāgara 34.69–73 and 56.212 for descriptions of khanyāvādī and bilavādī Pāśupata ascetics.

262 “the science of controlling the earth” (mahīvāda): Alexis SANDERSON suggested this emendation. Mahīvāda is not found in lists of magical sciences, but a synonym ketravāda (whose meaning is never explicitly stated) is mentioned in Śaiva sources among the mantravādas (see e.g. Śivadharmottara, Wellcome Institute for the History of Medicine, London, South Asian MS Collection, No. 16, f. 3r7–f. 3v3, and JRY 1.45.150–151a (NAK 5-4650, f. 161v3)). Most of the witnesses of Sαβ∞ have °mahāvāde (interpreted as a vocative by Ballāla at f. 26r1) which seems corrupt. Rasārava 1.44 gives a hierarchy of siddhis: khanya° (a variant found in witness M; the edition has khaga°), bila°, mantra° and rasa°. The emendation of mahīvāda to mantravāda would, however, be unmetrical. K2’s svarādidhātuvādāni, “the sciences of metals such as gold etc.”, for the whole pāda is noteworthy but most probably a scribal emendation.

263 Many of my informants told me that the practice of khecarīmudrā enables the yogin to go without food and water, a skill necessary for extended periods of yogābhyāsa; this is also stated at GŚN 65 (=HP 3.38), p.2 1.3, GhS 3.24, ŚS 3.93, 5.60 and by BERNARD (1982:68). An addition in the margin of S (f. 21r) quotes Yogasūtra 3.29 which suggests early origins for this idea: kahakūpe kutpipāsānivtti “[sayama] upon the hollow of the throat [brings about] the suppression of hunger and thirst”. Cf. The passage from the Suttanipāta (p. 138, vv. 716–718) cited on pp. 18–19 of the introduction and note 425 on extended samādhi.

Ballāla understands vratastha to mean “living as a brahmacārin ”, i.e. practising celibacy: guptedriyasyopasthasayama (f. 27r2).

264 This odd-sounding assertion probably means that the siddhis only arise as a result of the mental and physical practices which are focussed on the region between the eyebrows (cf. 1.66b). This emphasis on the importance of the region between the eyebrows contradicts 1.50a–55d, where the tongue is to be lengthened until externally it reaches the top of the skull (see note 245), suggesting that the two passages were not composed together.

At 2.22cd the somamaala is said to be between the eyebrows; this verse may be referring to that place.

265 “in the ether” (ākāśe): here ākāśa means the cavity above the soft palate. See page 28 of the introduction. Cf.JRY 4.2.157a, MVUT 21.2, 3.137–140, AM 67.1, GBS 23; see also WHITE 1996:240–242. The Khecarīvidyā’s subtle physiology does not include a system of bodily voids such as those found in some texts of Śaivism and hahayoga (on which see VASUDEVA 2004:263–271).

266 There is disagreement both between the witnesses of the KhV and between other hahayogic texts over whether or not the teeth should be clenched during the practice. Witnesses A and K3, and TŚBM 92 and 146 say that they should not; all the other witnesses, Mahopaniad 5.75 and ŚS 3.87 say that they should. Clenching the teeth is the preference of the more ascetic tradition—it is mentioned in the passages from the Pali canon cited in the introduction (pp.17–19) and is consistent with the ideas of effort and force implicit in the name hahayoga. Not clenching the teeth is favoured by the tantric tradition: cf. KMT 23.161c (see page 23 of my introduction). In instructions for physical postures to be adopted during sādhana (but not specifically connected with khecarīmudrā) Mgendratantra yogapāda 19c, Sarvajñānottaratantra yogapāda 12a (see VASUDEVA 2004:398 n.77) and JRY 4.2.683c instruct the sādhaka not to touch his teeth with his teeth.

267 “making the mouth [like] the hollow of a crow's beak” (kākacañcupua vaktra): during the practice fluid gathers in the mouth. By pushing out the lips into the shape of a bird's beak there is more room for fluid to collect. In the hahayogic practice of jālandharabandha (described at GŚN 62–63, HP 3.70–72, etc.) the throat is constricted by letting the head hang forward. The fluid dripping from the moon is thus diverted into the mouth and prevented from falling into the solar region at the stomach (hence the suitability of the name of the practice, jālandhara, which can be interpreted as a vddhi derivative from jaladhara, “holding water”). Jogpradīpakā 886–889 describes a jālandharabandha in which the tongue is placed in the middle of the trighaī, the three sets of orifices at the nostrils, eyes and ears. Vivekadarpa 10 mentions the kākīmukhī attitude in connection with jālandharabandha. Instructions to make the mouth like a bird's beak when practising khecarīmudrā are also to be found at JRY 4.2.157, KhV 3.25, GŚN 139, ŚS 3.85, GhS 3.66.

268 Ballāla (f. 27r10) glosses khecaratvam with devatvam: he does not equate khecaratva with the ability to fly. See note 113 in the introduction.

269 “all the magical powers” (siddhi†samayam†): the meaning of samayam here is not clear. It is tempting to emend °samayam to °santānam (cf. KhV 2.70a). However, Ballāla (f. 28v7) reads siddhasamaya (he glosses it with jhoigādivīrādibhūtapretādi and understands the verse to mean that the sādhakottama can quickly get control over all these beings), and I have found three instances of the compound siddhisamaya in Buddhist tantras—Guhyasamājatantra, prose section after 17.25:kāyasiddhisamayavajram (I am grateful to Harunaga ISAACSON for providing me with this reference); Savarodayatantra 18.30: siddhisamayasavara; Kayamāritantra p. 100: tathāgatasiddhisamaya.

A marginal note in S (f. 27v) adds that all these siddhis are described in the Dattātreyatantra and other texts (dattātreyatatrādau).

270 Pādukāsiddhi gives the yogin sandals that enable him to go wherever he wishes. Ballāla (f. 27v12) says that this siddhi is explained in the Nāgārjunatantra and the Tantrarāja, and that the sandals can be used to cross water and travel long distances. MaSa paala 30 (MS A f. 70v6–f. 71r8) describes pādukāsiddhi: the sādhaka is to make sandals out of various precious metals, go to a cremation-ground, drink alcohol and repeat a saptakūamantra one lakh times. The sandals will thus be empowered by the Yoginīs of the cremation-ground. Cf. Kulacūāmaitantra 6.25c–26b, ROBISON 1979:258. Tantrarājatantra paala 17 describes the mantras and effects of sixteen siddhis, including pādukā, khaga, vetāla, añjana, ceaka and yakiī.

271 MKSK paala 6 (pp.52–54) describes khagasiddhi: by means of mantras, an offering of his own blood, and, if possible, a human sacrifice (narabali), the sādhaka empowers a sword to guarantee him victory in any battle. Cf. Kulacūāmaitantra 6.26c–33d.

272 “power over zombies” (vetālasiddhi): Ballāla (f. 27v3) explains this siddhi as piśācavaśitvam, “control over ghouls”.MaSa paala 32 (MS A f. 74v8–f. 75r11) describes vetālasiddhi: the sādhaka should drink alcohol, repeat a saptakūamantra one lakh times and make a tarpaa offering of goat's blood. If performed correctly, a vetāla appears and becomes his lifelong servant. Cf. Kulacūāmaitantra 6.19a–25b.

273 Realgar (manaśilā) is red arsenic, an ingredient in elixirs: see e.g. KhV 4.9. Picumata 46.57 (NAK MS No. 3-370, f. 224v) includes manaśilā in a list of siddhis, and a Buddhist Kriyātantra, the Amoghapāśakalparāja, describes how manaśilā, when applied to the eyes, can make the wearer invisible and able to move in the ether: manaśilā añjana vā parijapya akiy añjayitvā tato ’ntarhito bhavati. ākāśena parikramati (pp.2–3). Cf. Kāmasūtra 7.2.46, in which it is said that if one coats one's hand with the faeces of a peacock that has eaten haritāla and/or manaśilā and touches something, it becomes invisible. I am grateful to Alexis SANDERSON for providing me with these references.

274 MaSa paala 35 (MS A f. 78r11–f. 81r9) gives instructions for añjanasiddhi: after a mantra-repetition and visualisation, various recipes are given for the preparation of the ointment (añjana) whose ingredients include herbs, honey and, in one concoction, mercury. By applying this ointment to the eyes, the sādhaka “sees everything” (sarva paśyati). Cf. MKSK paala 6 (p.55), Kulacūāmaitantra 6.34–39. Ballāla (f. 27v4) says that this siddhi has been described in the Nāgārjuna and Dattātreya Tantras.

275 Vivarasiddhi is similar to the bilasiddhi mentioned at I.68c. Ballāla (f. 27v4–6) glosses it with bhuvas tatsādhana and explains it as the ability to enter ponds, wells, tanks, caves and ditches, and retrieve treasure therefrom.

276 Ceakas and yakiīs are genie-like male and female servants respectively. Ballāla (f. 27v8) describes ceakasiddhi as parapreyakāritvam, “the power to enslave others”. He says that the best slave is the gaeśaceaka and gives his mantra. (µ’s reading for ceakam, kheakam, “shield”, may be original.) Yakiīs are usually associated with Kubera and can bestow wealth and sexual favours (see e.g. KSS 37.64–83, BKhP f 28v1 –f. 28v5). MaSa paala 34 (MS A f. 76r1– f. 78r10) describes yakiīsiddhi: by means of a trikūa and other mantras, and a visualisation of the goddess, the sādhaka gets yakiīmelaka. Ballāla (f 27v10–f 27r1) gives a yakiīsiddhimantra and says that according to the Nāgārjuna and Dattātreya Tantras there are thirty-two yakiīs.

The syntax of this list of siddhis is odd. 1.75 cd is a plural dvandva compound while 1.76ab lists its siddhis one by one. G omits 1.75 cd, suggesting that the line may be a later addition to the text. Some of the witnesses seem to have attempted to split 1.75cd into separate elements, but it is metrically impossible to alter °khadgavetāla° to khadgo vetālah.

277 The basic meaning of kalā is “a part”, especially “a sixteenth part of the moon” (e.g. Bhadārayakopaniad 1.5.14; see GONDA 1965:115–130). The moon waxes and wanes in periods of fifteen days; each day it gains or loses one kalā. The sixteenth kalā is the amtakalā (SSP 1.63; cf. CN 46) which never dies, even at the dark of the moon. (Some tantric texts add a seventeenth kalā; see e.g. 3.137, Jayaratha ad TĀ 5.6364, Parātrīśikāvivaraa 35.) Many texts also describe the kalās of the sun and of fire (e.g. KAT 6.37–40, SSP 1.63–65, GBS 89). The moon's association with soma and amrta has led to all of its kalās being thought of as containing amta, and it is in the sense of a store of amta that the word kalā is used in the Khecarīvidyā. Kalā can also mean “tongue” (e.g. HP 3.33) and, in tantric descriptions of the phonematic emanation of reality, “vowel” (TĀ 5.63–64; PADOUX 1990a:89–91). See also note 311; WHITE 1996:36–44.

278 The four aims of man are kāma, artha, dharma and moksa. See 2.3–6. Cf. KJN 5.31, 8.41.

279 I have been unable to find parallels of this list of kalās in other texts.

280 Thus moka (see 2.1) is equated with paramehīnām ādhipatyam, “dominion over the highest gods” and subordinated to the end described in the next verse: becoming Śiva, liberated while living.

281 The witnesses here are unanimous in reading dvādaśābdam, “for twelve years”, or corruptions thereof. One would, however, expect a word meaning “for twelve months” rather than “for twelve years” because of the passage at 2.10–17, in which are listed the rewards obtained each month from drinking the parāmta in the brahmarandhra over a period of a year, culminating in the attainment of Śivahood.

282 For saha savartate G has sadā saveito: rather than merely associating with the beings listed, the yogin is forever surrounded by them.

283 Cf. GBS 138: parcay jogī unman khelā ahanisi ichyā karai devatā syu melā, “the yogin in the paricaya state plays in unmanī day and night, and meets with deities at will”.

284 Jogpradīpakā 638–640, in a section on praveśana (see Khecarīvidyā 2.105), lists the rewards of placing the tongue at the brahmavivara: after one month the īs are purified, after two months the yogin hears the anāhata nāda (Hindī anahad nād), after three months the body has a divine radiance, after four months the yogin has long-distance hearing, after five months the mind becomes like that of a child and after six months the yogin assumes the form of Śiva.

285 Kedāra is located between the eyebrows in HP 3.23 and Darśanopaniad 4.48, but KhV 2.49–56 clearly indicates that the cūlitala is at the back of the head, above the nape of the neck (see also note 241). The description at 2.22 of a further set of kalās at the somamaala between the eyebrows confirms that the KhV's Kedāra is not located there. WHITE (1996:245–246) describes parallels between the site of the Himalayan shrine of Kedārnāth and the subtle body of hahayogic physiology.

286 On the connection between amta and Soma, see note 277, DASGUPTA 1976:250–1 and GONDA 1965: ch. 2.

Instead of vīravandite, “o you who are worshipped by the extreme adepts”, Sßγ have the aiśa sandhi form ’maravandite, “o you who are worshipped by the gods”. This is the only occurrence of an aiśa form in the KhV manuscripts being found in a correct form in µ. I suspect that this is because of an attempt to get rid of vīra°, a word that has strong connotations of left-hand tantrism. Cf 2.11od, where almost all the witnesses read sasthitā vīravandite; only G has sasthitāmaravandite (in which the sandhi is correct).

287 The names of the next sixteen kalās that are listed (eight at Kedāra, four at the somamaala, three at the khecaramaala and the first of the two at the rājadanta) match exactly the sixteen saumyakalās listed at Kulāravatantra 6.37–38 and the lunar kalās listed in a quotation from the Merutantra in the third taraga of the Puraścaryārava (p. 215; in this list Pui and Tui are transposed). In Amtānanda-nātha's Dīpikā on Yoginīhdaya pūjāsaketa 104–105 he lists sixteen saumyakalās: Amtā, Mānadā, Pūā, Pui, Prīti, Revatī, Hrīmatī, Śrī, Kānti, Sudhā, Jyotsnā, Haimavatī, Chāyā, Sampūritā, Rāmā and Śyāmā. MONIER-WILLIAMS (1988:s.vv.), at the entries for each of the names of the eight kalās here located at the khecaramaala, says that they are (in the same order) the names of the kalās of the moon as described in the Brahmapurāa, but I have been unable to locate any such passage in that work.

288 In this and subsequent descriptions of groups of kalās, it seems that the yogin should spend a month tasting the amta at each kalā because the rewards to be gained are obtained after the same number of months as kalās at that particular kalāsthāna.

289 The Orb of Soma (somamaala) probably refers to the moon: the names of the kalās here have particularly lunar connotations; furthermore, in the Kaivalyadhām edition of the GŚN the moon is called somamaala in verse 56 (KAIVALYADHĀM 1991:314). MKSG 11.997 mentions a mahācakra called Soma above the forehead. MVUT 16.13 and VS 4.41 locate the somamaala at the heart.

290 samāpibet: the reading samāviśet found in almost all the witnesses seems odd, particularly after samāveśya earlier in the line. I have thus adopted SANDERSON’s conjectural emendation samāpibet. One could understand samāviśet to mean “the yogin should enter [samādhi]” but there are no similar constructions elsewhere in the text. Ballāla (f. 32v9) understands samāviśet to mean that the yogin should remain with his tongue in place: praviśyaiva sthito bhavet.

291 I have conjectured yogī for the first word of 24c where G and the KhV manuscripts have devi. Nowhere else in the text does a vocative start a half-verse. The manuscripts of µ have devabhāsacatukoa (°kea J6), “a square of divine appearance”, for 24c which does not fit the context and is probably a corruption of the reading found in the other witnesses.

292 Witnesses µG, which often preserve original readings, have khecaramadhyagam, “in the middle of Khecara”, for khecaramaalam, but I have been unable to locate any other references to a place called Khecara in the body so have adopted the reading of the KhV manuscripts. Khecaramaala perhaps refers to the sun, in contrast to the lunar somamaala that has just been described.

293 “known as the Diamond Bulb” (vajrakandākhyam): I have found no references to a vajrakanda in the body in other works on yoga. RAK 156 mentions a plant called vajrakanda in a description of a mercurial preparation. Several works describe an egg-shaped kanda or kandayoni at the navel as the source of the 72,000 īs, e.g. GŚN 25 (=YCU 14c–15b), VS 2.11–12 (=TŚBM 58–59), ŚP 4307. On the analogy of this kanda, the vajrakanda may be a point of intersection or origin of īs. See also 2.49c–50b, 2.86 and note 448.

294 The reading found in various forms in the manuscripts of Sβγ, alakya sarvalekhakai, is interpreted by Ballāla (f. 33r10) to mean “imperceptible by the gods”, i.e. “invisible” (adarśanīya).

295 nāsikādho ’dharohordhvam: this is an emendation of the reading found in G and, in a corrupt form, in µ. The witnesses of Sαβ have nāsikādhottaro° or corrupt versions of it (γ has the nonsensical nāsikādyotaroādha). This form is the result of a double sandhi (“āra” sandhi according to Ballāla at f. 33v1) of nāsikādha+uttaro°. I have taken adharoha to mean both the upper and lower lips (cf. MONIER-WILLIAMS1988:19).

296 As we have seen in note 258 the rājadanta or “Royal Tooth” is the uvula, so its description as “below the nostrils and above the lips” is surprising and suggests that it means somewhere in the region of the front teeth. Presumably the description means that the rājadanta is on the same horizontal plane as the space below the nostrils and above the lips.

297 “the Base” (ādhāram): this is the ādhāra or mūlādhāra cakra of tantric and yogic physiology. See e.g. KJN 14.15–24b, KMT 13.37–52, CN 4, GŚN 18, YŚU 1.168.

298 I have found no parallels for this or any of the subsequent lists of kalās.

299 i.e. by means of the hahayogic mūlabandha. Ballāla (f. 34v1–2) explains it to be the forcing of breath into the head by sitting in padmāsana or siddhāsana, contracting the Base and repeating hu hu: ākucana tu padmasiddhāsanasthatve sati huh* u* kāreādhārakamala sakocya tatrasthavāyo p.havaśe nayanābhojavttau tu nābhimūlāt preritasya vāyo śirasy abhihananam /. Cf. GŚN 58–59, YKU 1.64, HP 3.60–68 etc. In his commentary on NT 7.30, Kemarāja describes a forerunner of this practice in which the contraction and expansion of the anus cause Kualinī to point upwards: cittaprāaikāgryea kandabhūmim avatabhya tanmūlam iti mattagandhasthāna śanair iti sakocavikāsābhyāsena śaktyunmeam upalakya pīayet yathā śaktir ūrdhvamukhaiva bhavati (see also NTU pp. 157–158). T. is repeated contraction and expansion is a feature of the Gheraasahitā’s aśvinīmudrā which is also said to awaken Kualinī (GhS 3.64).

300 “up to his skull” (brahmāakāvadhi): in the KhV brahmāa means skull. See note 257.

301 Ballāla (f. 35v7–8) explains pañcabhūtalaya as absorption into the subtle elements: yady api sthūlānā bhūtānā layo 〈’〉sabhavas tathāpi tanmātrāā lavarayahānā bījabhūtānā tatra tatra japenetadevatādhyānena ca laye tallayasyārthasiddhatvāt. In hahayogic texts, laya is both an aim of yoga (see KhV 3.48–52; AY 1.21– 98, p.5 ll.16–20, HP 4.3, 4.29–34 etc.) and a type of yoga itself (e.g. DYŚ 29–30 and 37–51, HP 4.103, YB 143, ŚP 4350–4363, VU 5.10). µ’s reading, pañcabhūtajaya labhet, preserves an older idea of mastery over the elements found in the Yogasūtra (3.43) and many tantric works (see VASUDEVA 2004:240–250). In his commentary on 3.65 (f. 97r68), Ballāla quotes a passage on bhūtajaya which he attributes to the Mahābhārata: bhārate paukare saptadaś ādhyāye nīlakah*kramea pādādi jānuparyata / jānvādi pāyvata / pāyvādi hdayāta / tato bhrūmadhyāta / tato mūrdhāta / cakrapacaka (pacaka] em.; paca S) pacaghaikāparyata mano dhārayato bījāni japata uktadevān dhyāyataś ca tattadbhūtajayo ’vaśya bhavati /. See also note 316.

302 “in the morning, in the evening and at midnight” (trikālābhyāsayogata) : this is the conventional meaning of trikāla and it is understood thus by Ballāla (f 36r1–2).

303 “the place of the penis” (ligasthānam) : the Svādhihāna lotus is located in the region of the penis in GŚN 22, ŚS 5.103, CN 14 etc. Likewise Ballāla puts it ligamūle, “at the root of the penis” at f. 36r7. I have thus adopted G's reading over µ’s incorrect nābhisthānam and the vague nābhisthānād adhah of the other witnesses.

304 As Ballāla notes (f. 37v8–10), this and the reward mentioned at 2.45b are presumably the reward described at 2.39.

305 “the Bamboo Staff” (veudaam) : I have not come across references to the veudaa in any other texts. Ballāla (f. 37r1) says that it is the lower part of the spine (phavaśākhyasya mūla) and equates it with the vajradaa described in YB 131. G's reading of vīā may be original: Tantrarājatantra 27.35 says that the ādaa is the spine—suumā prhavaśākhyavīādaasya madhyagā; YŚU 6.8 describes the ādaa as being in the region behind the anus and supporting the body (dehabht).

306 45c–48b appear to be a later addition to the text: 48c follows on directly from 45b.

307 Cf. GŚN 32, VS 2.27–28 etc.

308 The edition's reading of ravi prokta in 46a is attested only by B and is possibly a scribal emendation. It is tempting to adopt µ’s raver bāhu, taking it to mean “a ray of the sun” but I have found no parallels for this usage of bāhu The reading raver vāha found in a variety of forms in the other witnesses results in the unwanted repetition of vāha.

309 47a is puzzling and I suspect that the text is corrupt. I have found no parallel passages in other hahayogic texts. As Ballāla notes (f. 39r10), dhāraā can mean both fixing of the mind on a single object and fixing of the breath. (The two are linked: Vyāsa in his commentary to YS 2.12 states that mental dhāraā is brought about through breath-control; cf. HP 4.23.) I have interpreted this pāda with the former sense of dhāraā. It could also be interpreted with the latter sense, giving the meaning that the yogin is to inhale through the lunar channel but this would be somewhat redundant since the same is said in the next half-verse.

310 This lunar prāāyāma with its emphasis on inhalation through the Iā ī has no parallel in the manuals of hahayoga, in which the yogin is usually instructed to use alternate nostrils for inhalation (e.g. HP 2.7–10, ŚS 3.24–25, GhS 5.38–53). There is one technique in which the yogin is to use only one nostril for inhalation: sūryabhedana (HP 2.48–50, GhS 5.58–63); however it is the Pigalā ī which is to be used for inhalation and the Iā for exhalation.

311 The kalās situated in the lower part of the body total twelve (five at the Base, three at the Svādhihāna and four at the Bamboo Staff). This figure tallies with the descriptions of twelve kalās of the sun (which is situated in the lower part of the body in yogic physiology: see e.g. HP 3.76–81) found in SSP 1.64 and KAT 6.39. This may be coincidence: here the kalās are not said to have any connection with the sun while in the SSP and KAT passages the names of the kalās are explicitly solar. Moreover, no such neat correspondence can be made for the twenty-two kalās situated in the head. Indeed it is striking that the kalās in the head do not total sixteen or seventeen (see note 277). (Ballāla (f. 38r7) omits the four kalās at the somamaala and the single kalā above the brahmārgaladvāra to arrive at the scripturally prescribed total of seventeen candrakalās.)

312 I usually translate sudhā as “nectar”, amta as “amta”, parāmta as “great amta”, and paramāmta as “supreme amta”. I have chosen to translate parāmta here as “ultimate amta” because Śiva is now teaching the location of the highest store of amta in the head.

313 The Diamond Bulb (vajrakanda) has been described at 2.25c-29b. See note 293.

314 For yoginah, “yogins”, µ has yoginya, “yoginīs”. This may indicate a difference in doctrine between µ and the other witnesses, but could also be because of a scribal error.

315 On the cūlitala see note 241.

316 This description of five places in the head corresponds to descriptions of the qualities of the five elements to be meditated upon in the hahayogic dhāraā (e.g. GSN 155– 159, VS4. 4.1–15, DYŚ 220–242, ŚS 3.72–74, GhS 3.59–63, Śivasvarodaya 209–213; cf. MVUT 13.21c–13.53d, Mgendratantra Yogapāda 39–44; Ballāla (f 42(1)v) quotes similar passages from the Kulaprakāśatantra (see KAVIRĀJ 1972:143), the Śāradātilaka and the Mahākapilapañcarātra (see ibid.:484)). These elemental qualities (appearance, colour, shape, bīja etc.) have been imposed (with some differences) upon different sets of five physical locations in different schemata of esoteric physiology. Thus they appear in the CN's description of the lower five cakras at the perineum, the genital region, the navel, the heart and the throat (cf. ŚS 3.73–74); in the GŚN they are found at the heart, the throat, the palate, between the eyebrows and at the brahmarandhra; in the DYŚ they are in the regions between the anus and navel, at the navel, above the navel, between the navel and the eyebrows and above the eyebrows; here in the KhV the first four are at the cardinal directions in the head with the fifth above, in the centre. The order in which the elements are listed here is different from that found elsewhere. In the text from 49c to 58b and in its summary at 58c–59d the order is earth (pthivī), fire (sūrya), air (anila), water (jala) and ether (ākāša), in contrast with the usual order of earth, water, fire, air, ether. They are, however, positioned in their usual order as one circumambulates the head (albeit anticlockwise): starting at the forehead with earth, there is water at the left temple, fire at the back of the head, air at the right temple and ether on top.

317 At f 40v10 Ballāla likens the four ligas with the fifth in the middle and a store of cooling amta above to the four columns of a temple with the liga in the middle and a galantikā or kalaśa dripping water onto the liga from above: caturdiku galatikāstambhās tadupari pragalajjalakalaśa.

318 “with the moon above it” (ūrdhvacandram): the readings of µ and G (ūrdhver ūrdhva° and ūrdhvaradhra° respectively), although corrupt, suggest that ūrdhvacandram may not be the original reading. The Vairāapurāa locates a cakra called both ūrdhvarandhra and tālucakra above the sahasrāracakra (KAVIRĀJ 1987:51).

319 “perfect” (heyopādeyarahitam) : heyopādeyarahita literally means “free of those things which are to be rejected (heya) or cultivated (upādeya)”, i.e. free of any hierarchised duality. Ballāla (f 43r10) glosses heya with samsāra and upādeya with moka. The Mālinīvijayottaratantra opens with a statement of what is upādeya and what is heya (1.14c–17b): “Śiva, Śakti and Sovereigns of Mantra-regents, Mantras, Mantra-regents, and individual souls” are to be cultivated. “Impurity, karma, Māyā, the entire universe deriving from Māyā” are to be rejected (VASUDEVA'S translation (2004:151)).

320 “to obtain the ultimate substance” (paratattvopalabdhaye): here paratattva can be understood both physically and metaphysically. It is amta, the ultimate substance, beyond the five elements already mentioned, and it is the ultimate reality, the goal of many tantric and hahayogic practices (see e.g. KT 59.36, HP 4.37 and KhV 2.100c). That this practice is not entirely physical is indicated by phrases such as manasā saha at 64c and 65d, and unmanyā tatra sayogam at 67c. Most of the first adhyāya of the Amanaskayoga (vv. 21–98) is devoted to describing laya, by means of which the paratattva is obtained. (This first adhyāya is called layayoga when quoted from by later commentators; the second, which describes amanaskayoga, is called rājayoga (BOUY 1994:22, 69, 78)).

321 On Vāgīśā, “the goddess of speech”, see 2.11ocd.

322 “with her mouth upwards” (ūrdhvavaktrām): i.e. with the tip of the tongue pointing upwards in order to lick at the amta. G's reading, ūrdhvavaktre, suggests the “upper mouth” at the opening of the śakhinī nāī from which amta flows (cf p.10).

323 The pot of amta is a recurrent theme in Indian mythology. When the ocean of milk was churned by the gods and demons Dhanvantri appeared carrying a white pot (kamaalu) of amta (Mahābhārata 1.18). Four drops of amta fell from this pot at the sites of the triennial Kumbh Melā, “Pot Festival” (on this recent addition to the myth see LOCHTEFELD 2004). At SYM 21.7 Bhairava is to be visualised in the middle of the Umāmaheśvara cakra churning a pot (kalaśa) full of amta; at SYM 22.36 in a description of the fearsome Yoginīcakra, at the hub of the wheel the Yoginīs churn and drink from a white pot (kalaśa) full of amta. The inner shrine of the Nātha monastery at Caughera in Nepal contains a pot of amta (amtapātra) which is said to be the svarūp of Gorakhnāth (BOUILLIER1997:31–32).

324 Unmanī, “the supramental state”, is a common goal of tantric and hahayogic practices. At HP 4.3– 4 it is included in a list of synonyms of samādhi. It is also frequently mentioned by Hindī poets of the nirgua tradition (CALLE'WAERT & DE BEECK 1991:626).

325 “that consists of nāda and bindu” (nādabindumayam): in tantric works, nāda and bindu (often combined with kalā—see note 277) have several different meanings. In particular, they refer to places in the body (e.g. NT 7.29, KT 58.56, Vijñānabhairava 36–37) and describe corresponding stages in the manifestation of the phonetic universe (e.g. Śāradātilakatantra 1.7–8, 4.175; see PADOUX 1990a:86– 121). In some texts they are also listed among the six lakyas, “the six manifestations of Śiva as the ‘goals’, or ‘targets’, of yogic practice” (VASUDEVA 2004:253– 292). In the texts of hahayoga, nāda is usually the internal, “unstruck” (anāhata) sound heard during yoga practice (see e.g. HP 4.66–106, Nādabindūpaniad 31– 51). Meanwhile bindu is understood to be the amta secreted in the head, which the yogin must prevent from falling and being discharged as semen (HP 2.78, 3.42, SSP 2.13; but cf. ŚS 5.144 where nāda and bindu, together with śakti, are pīhas in the lotus of the forehead; cf. also BKhP f. 100v3: hatraya bhāle bidunādaśaktirūpa / tatphala janmātarasmti / viparītajihvayā nādadhyāna pāpanāśana / śaktau vāsanākaya, and MaSa 17.14–16, in which the viśuddhacakra and an unnamed cakra somewhere above viśuddha are said to be nādarūpaka and bindurūpaka respectively). It is with the usual hahahayogic meanings that Ballāla (f 44v3–4) understands nāda and bindu (cf. HPJ ad 4.1). As such, the compound nādabindu joins two unconnected concepts and his interpretation seems forced. I suspect that in hahayogic works the compound is used more as a catchphrase, harking back to its use in tantric texts and thereby adding esoteric gravitas (see e.g. HP 4.1, GBS 163, 181, 184 etc., YŚU6.70, GhS 6.12 and the Nādabindūpaniad, which, despite its title, concerns only the “unstruck” nāda and mentions bindu just once, at verse 50). On nāda, bindu and kalā see also KIEHNLE 1997:141. At HP 3.46 in the description of khecarīmudrā the yogin is said to eat beef and drink wine (see page 28 of the introduction). The jogī is said to drink vāruī at GBS 137. Cf. Rasārava 1.26, Rasendracūāmai 1.7–10. See also ROŞU 1997:413.

326 Ballāla (f. 45v4–7) takes this verse to be describing those entitled to teach and learn Khecarī yoga: the text is to be spoken by [a yogin] who has no desire for siddhi (na ki cit siddhim icchatā—he interprets siddhi here as śiyād dravyasevādiprāpti, “obtaining goods, service etc. from a pupil”!) to one who has attained the means of siddhi (siddhisopānam) but does not know this yoga.

327 The aiśa anacoluthon in this verse has been emended in G, or one of its antecedents.

328 Practices involving massaging the body with various physical secretions are alluded to fleetingly in many hahayogic texts (see the references in the notes to 2.75a–77b). Paala 27 of the Matsyendrasahitā (which is reproduced on page 154 of the appendices) describes several such techniques in detail, summarising them as “the ritual bath which is better than [bathing] at all the sacred bathing places” (sarvatīrthādhika snānam 27.1cd). (At MaSa 27.2 faeces, urine, menstrual blood, phlegm (? recaka) and semen (? sāraka) are said to be the gods Lokeśa, Keśava, Rudra, Īśa and Sadeśvara.) These practices corporealise the techniques of rasašāstra, alchemy. (On corporealisation see page 27.) The words used to describe the massaging of the body, lepana and mardana, are also used to describe saskāras in the process of fixing mercury (see e.g. RAK 80, 91, 150 etc. on lepana and 54, 89, 98 etc. on mardana). As with the substances to be rubbed into mercury in the alchemical saskāras, in MaSa paala 27 minerals and herbs are added to the fluids to be massaged into the body. MaSa 27.1 calls the knowledge of these practices ketrajñāna; at Rasārava 18.11, 18.15 and 18.19 the preparation of the human body for alchemical practice by the consumption of herbal preparations is called ketrlkaraa (see also WHITE 1996:265–273).

In verses 95 and 281c–282b of the RAK it is said that the urine and faeces of a man who eats certain herbal preparations (which do not include mercury) can transmute copper into gold (cf. ŚS 3.61 and DYŚ 197 quoted in note 336). At 146 it is said that by eating a preparation of calcined mercury, a man becomes sparśavedhī and his sweat can fix mercury.

The physical practices are attacked at SSP 6.90:

šakhakālanam antara rasanayā tālvohanāsārasam
vānter ucchadanam kavāam amarīpāna tathā kharparīm /
vīrya drāvitam ātmaja punar aho grāsa pralepa ca vā
ye kurvati jaās tu te na hi phala teā tu siddhāntajam
//

90b ucchadana] em.; ullaana Ed, uchuhana Edvl ⋄kharparīm ] em.; kharparī Ed 90c vīrya] em.; vīrya Ed grāsa pralepa ca vā] grāsaprada pañcadhā Edvl

“Those who practise emesis and enema [and] use the fluids from the palate, lips and nose with the tongue, who massage themselves with vomit, who practise kavāta (?), drink their own urine, use Kharparī (coryllium?), who use their semen after causing it to flow, and eat or massage [themselves with these fluids], are stupid and do not get the reward that is produced by the correct doctrine.” (Several verses towards the end of SSP paala 6 appear to be later additions to the text since they contradict earlier verses: see e.g. 6.13 where the avadhūta who drinks his own urine is praised.) Cf. AY 2.33 (= AP 8): ke cin mūtra pibanti svamalam…, “some drink urine, their own filth…”, and Rasārava 1.11c–12b: śukramūtrapuriāā yadi muktir nievanāt / ki na muktā mahādevi śvānaśūkarajātaya // “If liberation [comes] from using semen, urine and faeces, then why are dogs and pigs not liberated, o great goddess?”

329 Here Śiva teaches the physical locations of fire, the sun and the moon. As in the locations of the five elemental deities discussed in note 316, the system described here is different from that found in other tantric and hahayogic texts. In the texts of hahayoga, the sun and fire are combined and said to dwell in the region of the navel, consuming the amta that drips from the moon which is situated at the palate (see e.g. GŚN 133, HP 3.78, GhS 3.29).

In this verse, only µ has bhālamadhye, “in the middle of the forehead”, (cf bhālajam at 74b), which is almost certainly original in the light of both 2.75, where the amta that has dripped from the moon emerges from the nostrils, and 2.22, where the somamadala, i.e. the moon, is located between the eyebrows. Similarly, at SSP 2.21 the yogin is told to visualise a candramaala at the bhrūmadhyādhāra while ŚS 5.182 locates the moon at the Sahasrāra lotus at the top of the skull. The readings of the rest of the witnesses, which locate the moon at the palate, have probably originated through confusion with other texts, rather than through deliberate alteration.

330 On mathana, “churning”, see 1.57c–64d.

331 Ballāla (f. 46v1–3) describes how the orb of fire is to be awakened: dakahastasya madhyamāguhābhyā amaruvan nāsikāpue pūrayan recayaś (em.; pūraya*ecayaś S) ca vādayitvā paścād gāha pūrayed recayed ity eātra bhastrā (em.; bhasrā S) tayā suumāvahane sati tadadhihitavahner udbodhana bhavatīti. This is a variation of the bhastrā/bhastrikā prāāyāma described at YB 108–112, HP 2.59–67 and GhS 5.70–72 (in the almost identical YB and HP passages the practice is said to be kualībodhakam and bring about śarīrāgnivivardhanam, i.e. it awakens Kualinī and increases the bodily fire).

332 HP 3.48 describes how amta flows from the moon after it has been liquefied by the heat produced when the tongue enters the opening above the palate: jihvāpraveśa-sambhūtavahninotpādita khalu / candrāt sravati ya sāra sā syād amaravāruī //. Cf. 4.131cd, 134ab.

333 “in a vessel” (pātrea) : MaSa 40.8 (A f. 9Ov 2) says that the vessel used to hold the yogin's urine (amarī) should be made of gold or silver, or, if they are unavailable, copper or brass (kāsya). KJN 12.11-16 describes the different materials that can be used to make the pātra that holds the cāruka, the Kaula pañcāmta libation, which consists of faeces, urine, semen, blood and marrow (KJN 11.11); cf. 11 (29) p. 130, ll. 5–8 where the five jewels are said to be urine, semen, menstrual blood, faeces and phlegm (SANDERSON 1995:82)).

334 Cf. MaSam 27.9. A corrupt passage at HP 3.93–94 describes the amarolī technique: amarī ya piben nitya nasya kurvan (nasya kurvan] tasya kuryād Edvl) dine dine / vajrolīm abhyaset samyag amarolīti kathyate// abhyāsān nis cāndrī vibhūtyā saha miśrayet / dhārayed uttamāgeu divyadi prajāyate // “He who always drinks urine, [also] using it as a nasally administered substance, every day [and who] correctly performs vajrolī, [his practice] is called amarolī. He should mix with ash the lunar [fluid] that has emerged after practice and put it on his head; he gets divine sight.” (Translation by GOODALL & ISAACSON.) Brahmānanda (HPJ ad loc.) attributes this practice to Kāpālikas.

335 In hahayogic texts, īśuddhi, “purification of the channels of the body”, is usually said to arise by means of prāāyāma. See e.g. N 95, HP 2.19, GhS 5.2, 5.35, SSP 6.79.

336 “the essence of immortality produced at the anus and penis” (gudaligodgatamamarīrasam): cf. MaSa 27.6. MaSa paala 40 describes amarīsnāna in detail and calls the process kulācāra (40.1). Ballāla (f. 47r6–7) quotes a passage in this context which he attributes to “traditional teaching”: mūtrapurīayor alpatvaca / yallepāl lohasya svaratā gorakasyeva tadā vajrolyā sādhitaliganālo mūtrasyāgrimadhārā viarūpā tathātimā hīnaguā satyajya madhyamā balapradāghītvaivam eva madhyamam alpa mala ghītvāga mardayed iti / paraparopadeśāt /. For modern accounts of urine massage see SARASVATĪ 1991 (especially pp. 74–76) and ARMSTRONG 1994. At f. 47v2–4 Ballāla describes the amarī and ajarī kriyās in which the yogin is to consume faeces and urine respectively: ke cittu gudodgata kakāmtena saloya *dharārasai sasktya yad bhakaa sāmarīkriyā / tatphala nirāmayatva balavattva ceti / ligodgata kakāmtena saloyādharārasai sasktya yat pāna sājarī kriyā tatphala valītyāgādī(°tyāgādī°] em.; °tyādī° S)ty āhu / amarī hy amarakāriī / ajarī a jarākāriī /. In contrast, PARRY (1994:290) reports that present-day Aghorī ascetics call urine amarī and faeces bajarī. GBS 141 says that he who practises bajarī and amarī is Gorakhnāth's guru-brother. The KhV's description of amarīrasa from the guda and liga and Ballāla's amarī and ajarī kriyās suggest that the Aghorī’s coprophagy is more than just a combination of opposites in which “pollution becomes indistinguishable from purity” (PARRY 1994:264).

ŚS 3.61 teaches how through perfection of prāāyāma, the yogin's urine and faeces can create gold or make it invisible: vimūtralepane svaram adśyakaraa tathā. Cf. DYŚ 197: malamūtrapralepena lohādīnā suvaratā /. See also Rasārava 18.28–29b.

HP 3.90 describes the sahajolī variant of vajrolīmudrā: sahajoliś cāmarolir vajrolyā eva bhedata / jale subhasma nikipya dagdhagomayasabhavam // vajrolīmaithunād ūrdhva strīpuso svāgalepanam / āsīnayo sukhenaiva muktavyāpārayo kaāt// . “Sahajolī and amarolī are types of vajrolī.[The yogin] should put good ash made from burnt cow-dung in water. Straight after sexual intercourse using vajrolī, it should be rubbed on the bodies of the man and woman, [when they are] sitting happily, free of activity”. It seems likely that this passage has been redacted to conceal a practice in which the combined sexual fluids of the yogin and his consort are smeared on the body. MaSam 40.48 describes a similar technique to be practised after intercourse although here it is only semen (mixed with gold, camphor, saffron and such like) that is to be smeared on the body: tad vīrya svarakarpūrakukumādiviloitam / svadeha mardayet kāntiś candravat saprajāyate. Jogpradīpakā 677–683 describes the varaaka mudrā which it says is also known as amarolī. The yogin is to drink three handfuls of urine first thing in the morning before mixing his urine with nirguī, bhagaro (=Skt. bhgarāja MCGREGOR 1995:s.v.), muī and giloī (= Skt. guūcī ibid.:s.v.) and smearing the mixture on his body.

The siddha Kararipa added his “own water” to a potion and it became “as the essence of the alchemists” (ROBINSON 1979:88–9).

The siddha Caparipa gave a child magical powers: “From his penis came the power to transform things into gold. From his anus came the elixir of immortality” (ibid.:206–7).

337 “the amta from the armpits” (kakāmtam) : the reading kalāmtam found in µ may be original. Śiva has described amtakalās at the anus and penis (2.32 and 2.40) but not at the armpits and MaSa paala 27 does not mention agamardana with sweat. However many hahayogic texts do teach that the sweat produced through yogic exertion should be rubbed into the body (e.g. GŚN 53, ŚS 3.46, HP 2.13, Dhyānabindūpaniad 70–72, DYŚ 148) and it may be because of this idea that the reading kakāmtam supplanted kalāmtam. ŚS 3.46 adds the reason for the practice: anyathā vigrahe dhātur nao bhavati yogina, “otherwise the basic constituents in the body of the yogin are destroyed”. Cf. Ballāla (f 95r8): evam saniyamaprāāyāme jāyamānasya dehe svedasya mardana hastābhyā kārya na tu vastreāpalāpa / lāghava balanāś;anāt /. “The sweat produced when prāāyāma is practised in this way, [i.e.] according to the rules, should be rubbed into the body with the hands, not wiped away with a cloth. Otherwise suppleness and strength are lost.” Like lepana and mardana, svedana is an alchemical saskāra (see Rasāravakalpa 98, 368–369 etc.).

338 “with fluid from the lower lip” (cādharārasai) : the feminine form adharā for adhara is probably metri causa. MONIER-WILLIAMS (1988:19) does report that adharā can mean “Pudendum Muliebre” but such a meaning is unlikely here. Ballāla (f. 47r10) takes the plural °rasai to indicate that fluid from the lips, tongue and nostrils should be used.

AY 2.33 castigates those who rub saliva into their bodies: … atha tanau ke cid ujjhanti lālām… naiteā dehasiddhir vigatanijamanorājayogād te syāt// Cf. SSP 6.90 (quoted in note 32. See WHITE 1996:311-2 for legends describing the initiatory and magical powers of yogins’ saliva.

339 I have found no description of this practice in any other text. Ballāla (f 47v8–10) identifies it as a supplementary practice to that described in 1.45a but he seems mistaken: at 1.45a it is the tālumūla which is to be rubbed and then all the impurity (mala) is to be cleansed. Here a potent “great fluid” (mahādrava; but n.b. µ’s reading madadrava, “intoxicating fluid”) is produced at the jihvāmūla.

340 “[the yogin] should push aside” (sphoayet): MONIER-WILLIAMS (1988:1270) gives “to push aside (a bolt)” as one of the meanings of the causative of the root sphut. Ballāla (f. 48r3–7) takes this verse to refer to the practice of ana (see note 236).

341 I have taken verses 80 and 81 to be summarising the practice of khecarīmudrā (un-like Ballāla who takes them with 78a–79d at f. 48r1–9). Verse 80 describes the process of inserting the tongue into the region above the palate. The tongue is to be pushed upwards (from its underside) while the uvula is to be brought forward thus making it easier for the tip of the tongue to reach the opening behind it (see note 229). The root k normally has a sense of “pull” or “draw” but if one were to pull the tongue upwards with the fingers of the right hand, the uvula would be inaccessible to the fingers of the left hand. I have thus taken utkya rasanām ūrdhvam to mean that the tongue is to be pushed upwards (as was demonstrated to me by several of my informants).

342 I am here following Ballāla's interpretation of ūrdhvavaktram as meaning lambikordhvakramam, “going above the uvula” (f. 48r8). Alternatively it could mean “the upper mouth”: see 3.23b and note 420.

343 “at the kalās” (candrāše): a part of the moon (candrāśa) is a kalā (see note 277).

The kalās referred to here are the three at the Diamond Bulb (vajrakanda—see 2.25c–29b and note 293) which is said to be the place of Śiva at 2.49c–50b (cf HP 1.48). For candrāše, μ has vajrātyo and G vajrāte. G's vajrāte may be the original reading, referring to the top of the vajrakanda.

344 In the Khecarīvidyā, trikūa, “the Three-peaked Mountain”, is located between the eyebrows. See note 259.

345 For similar accounts of curing physical afflictions by means of hahayogic practices, see HP upadeša 5, YB 102–112 and DU 6:25-30b; cf. MKSG 11.985. ŚP 4508– 4513 describes doopasargacikitsā by means of visualisation.

346 I have found no parallels for this usage of bhaa and naa. The usual meaning of bhaa is “mercenary” or “warrior” and that of naa is “actor” or “dancer” (MONIER-WILLIAMS 1988:s.vv.). The terms may thus refer to the different types of sādhaka that are afflicted by the problems listed. In Hindī, bhat can mean “misfortune, curse” (MCGREGOR 1995:757) while the Sanskrit root na can mean “to hurt or injure” (MONIER-WILLIAMS 1988:525). A Buddhist vihāra was established near Mathura by two brothers called Naa and Bhaa (Paś;upradānāvadāna, Divyāvadāna No.26, p.349; see also ibid. pp.356 and 385, EDGERTON s.v. naabhaikā, BÖHTLINGK and ROTH s.v. naa. I am grateful to Peter Wyzlic for supplying me with these references.) G, S and most of α and β have haa or haha for bhaa. I have adopted bhaa over haa/haha for three reasons: firstly, bhaa is found in both µ and γ; secondly, the use of the word haha to describe a system of practices was only just beginning at the time of the Khecarīvidyā’s composition and is not attested elsewhere in the text; and, thirdly, the pairing of haha with naa seems unlikely. Witness K5 lends weight to the idea that haha is a later emendation: at 84a and 99c it has bhaa, corrected to haa in the margin. Perhaps the first description of a systematised hahayoga named as such is to be found in the Dattātreyayogaśāstra (17–19 and 57–62) in which the term refers specifically to the practice often mudrās. The DYŚ is quoted extensively in the Śārgadharapaddhati (25 ślokas between ŚP 4376 and 4460) and was thus composed before 1363 CE.

In his commentary on naa, Ballāla devotes five folios (f 48v8–f 53r7) to quotations from various texts about akādinibaddharasādi, “the dramatic sentiments etc. involved in the various types of drama”. The Khecarīvidyā’s naabhedas are physical manifestations of these sentiments. When they arise, actors are unfit for acting: teu jāteu nartanayogyā naā na bhavati (f. 53r8). This is relevant to yogic practice because the sense organs are like the naas: vastutas tu svasvavyāpāre nartanašilānām naānām ivedriyāā netrādīnā bhedā bhedakā naabhedā ity ucyate (f. 53r7). Concerning haha (S's reading for bhaa), Ballāla (f 53r9–f 53v1) writes that the four manifestations of haha given in 83cd are proof of success in hahayoga (!): ete hahasya yogasya pratyayā hata siddha iti pratīti janayati.

347 “drying up of the body” (agaśoa) : at Kubjikāmatatantra 23.160 a practice similar to the hahayogic khecarīmudrā is said to get rid of ś;osa, dāha (cf. KhV 2.88d) and vaivara (cf. KhV 2.87cd).

348 “sloth induced by hunger” (kudhālasyam) : in order for the varieties of bhaa to total four, kudhālasya must be taken as a single entity. I have chosen to take it as a tatpurua samāsa; Ballāla (f. 53r9–10 and f. 53r8–9) understands it to be a dvandva samāsa with the meaning “hunger and sloth”.

349 “the essence of immortality” (amarīrasam): see 2.76cd.

350 i.e. agaśoa, “dryness of the body”, (2.83c) is cured.

351 i.e. it should be done every four hours: daśamadaśamaghaikāyām (Ballāla f 53v6). A ghaikā is 24 minutes.

352 See 2.25c–29b, 2.49c–50b and note 293 for descriptions of the Diamond Bulb (vajrakanda).

353 “amta [from the anus and penis]” (amarīm): see 2.76cd.

354 “trembling of the body” (agavepa): I have adopted G's agavepa to avoid repetition of agaśoa from 2.83c.

355 “dizziness” (bhrāntis) : bhrānti usually means ignorance (see e.g. KJN 5.1). In the context here, however, it must refer to a more mundane physical affliction. Ballāla (f. 54v2) glosses it with mānasī viparītadhī, “mental perversity”.

356 “high fever” (mahājvara): Ballāla says (f. 54v7) that mahājvara cannot be cured by doctors (bhiagbhir acikitsya) and adds that doctors’ medicines are no use in curing any of the problems of haha and naa: hahanaabhedeu bhiagauadha na calati.

357 Ballāla (f.54v9) takes the tathaiva ca that follows netrāndhatvam, “blindness”, to imply bādhiryam, “deafness”.

358 This is the practice described at 2.32c–39d. Only G has mūlādhārāt suumnāyā at 2.92a; the other witnesses have variants of svamūlāt śvāsasabhinnām, “from her base, together with the breath”.

359 This technique involving internal sounds is similar to the hahyogic nādānusandhāna (see e.g. SSP 6.91, HP 4.66106, NBU 31–51, VS 3.39–40, GhS 5.73–76), by means of which samādhi is realised (HP 4.81). Here, µ reads jalanāda for mahānāda. µs reading may be original: in a passage which is found at both HP 4.83–89 and NBU 32–38 it is said that in the beginning of the practice one of the sounds that arises is that of jaladhi, “the ocean” (HP 4.85a). On the other hand, HP 4.84ab reads śrūyate prathamābhyāse nādo nānāvidho mahān, “in the first [stage of the] practice a great sound of many kinds is heard”. HR 2.148 connects khecarīmudrā with the internal nāda. For a survey of descriptions of the technique of nāda and lists of the internal sounds found in tantric works see VASUDEVA (2004:273–280).

360 “in his ears” (karābhyām) : -ābhyām is sometimes used for the locative and genitive dual (-ayo) in Śaiva tantric works. See e.g. Svacchandatantra 2.231 Kemarāja adloc., JRY 3.38.158c and JRY 3 Yoginīsacaraprakaraa 1.63ab, 1.64. I am grateful to Alexis SANDERSON for providing me with these references.

361 “the sound of the roar of a great elephant” (mahāgajaravadhvanim): in the lists of the various sounds heard during nādānusandhāna given in hahayogic texts (see note 359 for references), no animal sounds are mentioned.

362 “the sound of Brahmā” (brahmanādam): Ballāla (f. 56r1–2) offers two explanations of brahmanāda: firstly he takes brahma° to mean bhat and thus brahmanāda is the same as the mahānāda of 93b; secondly brahmanāda is the anāhata “unstruck”, nāda that is the focus of nādānusandhāna (see note 359). µ has sihanādam, “the sound of a lion”, which may be original. The small whistle worn on a thread around the neck by Nāthas is called sihanāda. BRIGGS (1938:11), however, reports that the Yogīs understand it to be called thus because ideally the whistle is made of g, “(deer-)horn”. Cf. SSP 5.15a where it is called śgī-nāda, with a variant reading (from MALLIK’s edition) of sihanāda.

363 The sound of thunder, meghanāda, is given as one of the anāhata sounds at GhS 5.75 and VS 3.40. Aghora, “not terrific”, is a name of Śiva and of one of his most important mantras (see e.g. Pāśupatasūtra 3.21–26, KMT paala 9 and the Sis[sic] Purā (BAATHVĀL 1960:236–237), a work ascribed to Gorakhnāth, in which Aghor is said to be the best mantra).

364 “knowing all the categories of reality” (sarvatattvajñah): Ballāla (f. 57v2–8) mentions four systems of tattvas: that described in the Nārāyaayogasūtravtti in which there are two types of tattva, jaa and ajaa, corresponding to the prakti and purua of Sākhya; a śākta system of twenty-five tattvas; a system said to be found in the Śaivāgamas comprising fifty tattvas, including the twenty-five just mentioned; and the (presumably twenty-five) tattvas described by Kapila in the Bhāgavata[purāa]. Ballāla adds that the system of fifty tattvas found in the Śaivāgamas has been described by him in the Yogaratnākaragrantha. GHAROTE & BEDEKAR (1989:208) list two manuscripts of works entitled Yogaratnākara but they are ascribed to Viśveśvarānanda and Rāmānandayogin.

365 “into this peaceful supreme reality” (śānte pare tattve): cf. 2.63d. Ballāla (f. 57r10 – f. 57v1) says that this para tattvam is the state reached by means of the four mahā-vākyas of the Upaniads.

366 As Ballāla notes at f. 58r8–10, it is surprising to find cālana, “loosening”, named as one of the four stages when in the first paala's description of the practice cālana is only mentioned in passing (1.49) and not by name. The cutting of the frenum, however, is discussed in some detail (1.46–48) and one might expect chedana to be the first stage. In most of the other texts in which the practice is taught (e.g. GhS 1.28–31, 3.21, Hahapradīpikā (long recension) 5.147–151 (H2 f. 100r7–f. 100v5)), cālana is given much more emphasis than it is in the first paala of the Khecarīvidyā. In the Siddhasiddhāntapaddhati (2.19, 6.84) the tongue is to be lengthened by means of cālana; chedana is not mentioned. This suggests that KhV 1.44c–77b and 2.101c–105d were not composed together.

It seems likely that cālana here does not refer simply to the stretching of the tongue. Commenting on 1.49, Ballāla (f. 19v3–9) quotes YB 91–98 for a description of the cloth used to take hold of the tongue when practising cālana. He notes that the passage comes in the description of a mudrā for arousing Kualinī, the śakticālanamudrā. Nowhere in the YB passage is it stated where the cloth is to be applied. The Hindī translation of the text supplies nābhi, “the navel” as the location. Similarly, in the description of śakticālana found in the Gheraasahitā, a much later text, the cloth is to be wrapped around the nābhi (3.43). It is hard to imagine how such a practice could be performed. It is probably because Kualinī is located in the lower part of the body that the practice is thought to be carried out there too. (Another description of śakticālana at ŚS 4.105–111 says that it is to be done by means of the apānavāyu; see also ŚS 5.11 and YB 124a; Satyānanda SARASVATĪ (1993:38 5–6) says that nauli, churning of the stomach, should be used.) The description of the cloth at YB 91–92 is found in the HP's description of śakticālana at 3.109 without any instructions as to what to do with it. Brahmānanda (HPJ ad loc.) takes the description to be of the internal kanda above which Kualinī sleeps. Perhaps the earliest reference (pre-1450 CE; see BOUY 1994:40) to the hahayogic śakticālana is found in a text called the Gorakaśataka which is an unedited work, found in only four manuscripts, different from the more popular text of the same name (which is available in several editions; on the different Gorakaśatakas see note 9 in the introduction). BOUY (loc. cit.) has noted that the first eighty verses of the first chapter of the Yogakualyupaniad (whose second chapter is taken from the Khecarīvidyā’s first patala) are taken from this unedited Gorakasahitā. YKU 1.7–8 states that there are two methods of śakticālana: a technique called sarasvatīcālana, and prāāyāma. 1.9–18 describe sarasvatīcālana. Again the place where the cloth is to be applied is never explicitly stated. The wise yogin is to wrap it around tannāīm (1.11). The sarasvatī nāī ends at the tip of the tongue (VS 2.37, DU 4.21, ŚP 4311) and, as we have seen (KhV 1.49), vāgīśvarī, “the goddess of speech”, i.e. Sarasvatī, has her abode at the tongue. This leads me to believe that śakticālana is performed by wrapping a cloth around the tongue, not the stomach. The Haharatnāvalī (c. 17th century) confirms this and seems to preserve an understanding of śakticālana that had already been lost before the composition of other, older texts such as the Śivasahitā. HR 2.118–127 describes the practice in detail and states explicitly that the cloth is to be wrapped around the tongue. Ballāla connects the pulling of the tongue with the awakening of Kualinī in his commentary to 2.40–42 (f. 36v4–5): vastraveitajihvācālanena ca śakti prabodhya…, “awakening [Kualinī-]śakti by moving the tongue wrapped in a cloth…” and also at f. 37v3–4where he says that Kualinī is to be awakened āsanakubhakarasanācālanamudrādinā, “by āsana, breath-retention, moving the tongue, mudrā etc.”. Touching the palate with the tongue is said to bring about immediate upward movement of the breath (which is the yogic forerunner of the awakening of Kualinī) at Kiraatantra 59.3 5 (see page 20). Cf the practice shown to me by Dr. Tripāthī described in note 236.

367 The witnesses here appear to be corrupt. They all have cālana, “loosening”, as the first stage, mathana, “churning”, second, pāna, “drinking”, third and praveśana/praveśaka, “insertion”, fourth (except G, which has pramelanam fourth). This presents two problems. Firstly, praveśana/praveśaka needs to precede pāna—the tongue must be inserted into the cavity above the palate before amta can be drunk. (Ballāla notes this at two places (f. 58r4–7 and f. 58v1–2) and gives two conflicting explanations. At first he says that after mathana the upper kalās start to produce amta and thus there is an intermediate pāna before that which follows praveśana. At the second instance he employs the Mīmāsakas’ maxim that the order of words is some-times subordinate to the order of their meaning: śabdakramād arthakramasya kva cid balavattvāt) Secondly, in 102c–105d these stages are elaborated. No mention is made of pāna but a stage called bhedana is described between cālana and mathana. I have thus conjecturally emended 101d–102a from dvitīya mathana bhavet // ttīya pānam uddiam to dvitīya bhedana bhavet // ttīya mathana śastam. 5.125–6 in the long version of the Hahapradīpikā (witness H2) describes the khecarī technique as having six ancillaries (aaga) : chedana, cālana, dohana, pāigharaa, praveśa and mantrasādhana. The six ancillaries are described in detail over the next 93 verses. The Jogpradīpakā names six ancillaries of khecarīmudrā at verse 611, but has mathana for H2’s igharaa.

368 On “loosening” (cālana), see 1.45.

369 On “piercing” (bhedanam), see 1.56. Cf. MKSG 22.971 and 985. NT 7.29 locates the fourth of six cakras at the palate and calls it bhedana.

370 On “churning” (mathanam), see 1.57c–64d. This passage (103c–104d) is corrupt. G omits 104cd while µ omits bhedanam and mathanam in 103cd and has tavadati smaimagetatunā priye at 104ab. I have been unable to conjecture a suitable emendation but the meaning of the passage is clear.

371 “into the ether” (ākāśe) : on this use of ākāśa to mean the hollow above the palate see note 265.

372 Detailed instructions on praveśana are given at Jogpradīpakā 635–640.

373 “By breaking the bolt of Brahmā” (brahmārgalaprabhedena): for prabhedena all the witnesses except µ read °praveśena (N has praveśe tālumūlena). The idea of insertion is also present in jihvāsakramaena (which I have translated with “inserting the tongue”) so °praveśena is redundant. I have thus adopted µ’s °prabhedena and take the two pādas to be referring to bhedana and praveśana respectively.

374 “a condition of bliss” (ānandabhāvatvam): Ballāla (f 59r4) quotes (without attribution) the following to explain ānanda: yathā ratau yathā ca miabhojane yathā suuptau iti, “like [the feeling experienced] in love-making, eating sweets and deep sleep”.

375 “a decrease in sleep” (nidrāhāni) : Khecarīmudrā is said to remove the need for sleep at HP 3.38. In the Haharatnāvalī (f. 5v5) the adept is described as tyaktanidra, “not sleeping”.

376 “social intercourse” (sagamam) : cf. Yogasūtra 2.40: śaucāt svāgajugupsā parair asasargah, “from purification [arises] disgust for one's own body [and] not mixing with others”. Forsaking company (janasagavivarjana) is said to lead to perfection of yoga at HR 1.78. Both µ and G read sagamam here while most of the other witnesses have sagame. Ballāla (f. 59r8) understands sagame to mean amtasthānajihvāgrasayoge, “on the conjunction of the tip of the tongue and the place of amta”.

377 “food-consumption” (bhojanam) : in the texts of hahayoga and amongst today's hahayogins there are two different attitudes towards food consumption. As a result of success in yoga, the yogin either eats very little (e.g. HP 4.75) or he can eat as little or as much as he likes without any effect (e.g. DYŚ 157). Before attaining siddhi, however, the aspirant must curb his appetite (e.g. HP 1.15, ŚS 3.20, GhS 5.16) but he should not fast (e.g. GhS 5.31, ŚS 3.36). Cf. Bhagavadgītā 6.16–17.

378 “With his seed turned upwards” (ūrdhvaretā) : this is the only mention of semenretention in the text. Other hahayogic texts put much more emphasis on khecarīmudrā’s usefulness in preventing the loss of semen (see e.g. GŚN 69, which is reproduced at HP 3.41, and on page 29 of the introduction).

379 “endowed with the [eight] powers whose first is minuteness” (aimādiguānvita) : the locus classicus for these eight siddhis is Vyāsa ad Yogasūtra 3.44: aimā, “minuteness”, laghimā, “weightlessness”, mahimā, “hugeness”, prāpti, “the ability to reach anywhere at will”, prākāmya, “the ability to do what one wants”, vaśitva, “control over elements and animals”, īśitva, “sovereignty” and kāmāvasāyitva, “effecting one's desires”. Ballāla (f. 59v3–4) gives a list which has garimā, “heaviness”, in place of Vyāsa's kāmāvasāyitva. MaSa 18.36c–37b substitutes garimā for Vyāsa's mahimā. VASUDEVA (2004:364–365) translates Kemarāja's interpretation of the eight siddhis (or guas) as given in his Svacchandatantroddyota ad 10.1073 and adduces parallels from other tantric Śaiva works.

380 Śrī, “splendour”, is a name of Lakmī, the consort of Viu. Ballāla calls her Yogīsā (f. 60r3).

381 Vāgīśā, “the goddess of speech”, is a name of Sarasvatī, the consort of Brahmā. See 1.49 and note 366.

382 “the fetter of death” (bandhamtyu) : this is the fraenum linguae or frenum, the binding tendon at the root of the tongue. It is called bandhamtyu, “the fetter of death” because it ties down the tongue, preventing it from reaching amta, “non-death”. See 1.46 and note 230. One would expect this compound to be mtyubandha. Ballāla makes no comment on the odd order of its elements.

383 “o mistress of the host” (gaāmbike): the host (gaa) is Śiva's troop of attendants.

384 This is the area in the middle of the skull described at 2.57.

385 “absorption in it” (tallayam): Ballāla (f. 61r1) understands tallayam to mean either tatra sthāne layam, “absorption at the place of Śambhu”, or tasya manaso layam, “absorption of the mind”. I have taken tat to refer to unmanī.

386 “with [inner] vision” (dśā): it is of course impossible to look at the tip of the tongue when it is in the cavity above the palate so we must assume some sort of internal “sight”. Ballāla (f. 61r4) glosses dśā with atadyā. After a passage on laya at HP 4.23–34, we hear of the śāmbhavīmudrā (which brings about the same result as khecarīmudrāHP 4.37—and is called khecarīmudrā in one manuscript of the HP and in a quotation of the passage in the BKhP at f. 72r2) in which the yogin is to dissolve his mind and breath in the internal lakya: antarlakyavilīnacittapavana (4.37a). SSP 2.26–27 describes four antarlakyas. In the HP and SSP, however, there is no mention of “sight”, as such; only forms from √lak are used. Cf. ŚS 5.37; MaSa 18.29 mentions a laka at the forehead. It may be that G's tadā is the original reading. The reading rasān found in γ is perhaps inspired by the idea that the tongue tastes different flavours during the practice, an idea found in many other texts that describe the practice (see page 22 of the introduction), but not in the Khecarīvidyā.

387 “perfect” (heyopādeyavarjitam): on this adjective see note 319.

388 In contrast with the rest of the Khecarīvidyā, 2.107–115 fits the first of the two yogic paradigms described at ŚP 4365a–4371b in which the yogin is to raise his mind and breath by way of the central channel and cause bindu to enter the void. Among other rewards, he becomes ūrdhvaretā—his seed turns upwards. Amta, Kualinī and cakras are not mentioned. On the two paradigms and their attempted synthesis in hahayogic works, see pages 28 to 30 of the introduction.

389 “The Ethereal Gagā” (ākāśagagā): the homologue of the Gagā in hahayogic physiology is the Iā ī (see KhV 3.10). However, Iā only goes as far as the left nostril (TŚBM 70, VS 2.39) and is never said to reach the cranial vault (ākāśa—see note 265). It is thus unlikely to be the referent of ākāśagagā. On the macrocosmic level there is an ideal homologue of this ākāśaga in the high Himālaya: the Ākāśa Gagā flows from Tapovan, above Gaumukh, the glacial source of the Gagā.

390 “he instantly becomes a master poet” (kavitva labhate kaāt): Sarasvatī (see note 235) bestows kavitva. At GŚN 147 ( ŚS 3.84) the yogin is said to become a kavi by pressing the tongue against the rājadanta, drinking [amta], and meditating on the goddess that consists of amta (amtamayī devīm). Cf SYM paala 12 in which the sādhaka attains kavitva by visualising the goddess Parā (who is associated with Sarasvatī: SANDERSON 1990:43-51) as pouring nectar into his mouth.

391 Meditation on Laksmī bestows kingship. Cf. MaSa 34.58 where the the rājyalak mantra is said to make the Kaula practitioner a king.

392 “five innate constituents” (sahajā): I have not come across any parallels of this usage of sahajā nor a similar set of innate physical constituents in any other text.

393 “which embodies the supreme” (paramātmake) : this description of the body is odd. It is tempting to take paramātmake as a vocative addressed to the goddess (wrongly written for paramātmike) but such a usage is not attested elsewhere.

394 “through the fall of the father” (pitkayāt): witnesses µ and G have pitkaāt and parikaye respectively here, neither of which seems better than pitkayāt which is found in Sαβ. This unusual compound is glossed by Ballāla (f. 62r7) with pitśarīrāt which has then been altered in the margin by a later hand to pitvīryāt. Pitkayāt has a disparaging sense to it and Alexis SANDERSON has suggested that it may be some sort of yogic slang, implying a condemnation of householders who do not retain their semen. In Āyurvedic works, kaya refers to the decline of a bodily element (dhātu): see MEULENBELD 1974:458–9.

395 In the ninth month according to Ballāla (f. 62r8).

396 “he should insert” (viśet) : viśet is being used here with a causative sense; in 124b it may be taken as indicative or causative.

397 In 123a–124b, the gender of sahajā/sahaja is somewhat confused throughout the witnesses. In 121c–I22d it takes the gender of its referent and I have kept these genders in 123 a–124b. No other witness does the same but it is the only way I can see of being consistent.

398 “o Lady of the Kula” (kuleśvari): Ballāla (f. 63r1) understands kuleśvari to mean the “Mistress of Kualinī”— kulā kualinī tasyā īśvari niyatre. In Kaula tantric works, Kulešvarī is the highest goddess, the consort of Kuleśvara. See also note 6.

399 “insert” (praviśya): for similar instances of √viś having an indicative form and causative sense see 2.123d and 3.3a. Ballāla (f. 63r10) also understands it thus, glossing praviśya with praveśayitvā and praviśet (3.3a) with praveśayet (f. 63v1). In this passage (3.1–4) the subjects of the verbs are not clear. This is indicated by the confusion among the witnesses over whether Kudalinī is the object or subject in verse 1. I have chosen to adopt the readings of µ and G in which she is the object of praviśya (the use of praviśya with a causative sense adds to the subject/object confusion). I thus take yogī to be the subject of all the verbs in this passage. It is tempting to take Kualinī as the subject of pītvā and viśrāmya in 4a (cf. ŚCN 53a where Kualinī drinks the amta herself; at KhV 3.41c the tongue (vāgīśī) rests in the amta) but the yogin is clearly the subject of vibhāvayet and even in aiśa Sanskrit absolutives and main verbs usually share a subject.

400 Ballāla understands 3.1–31 to be an expansion of the description of the five sahajās given at 2.120–124. Thus Kualinī is described at verse 1, Suumā at 8, the tongue at 16, the palate at 23 and the brahmasthāna at 28. This somewhat forced schema may be due to the corrupt reading pacamam found at 3.24a in all the witnesses except µ and G (which have the correct pavanam). At f. 71r3 he takes pacamam to refer indirectly to the fifth sahajā, glossing it with bindu sthānagalitam.

401 “the bolt of Śiva's door” (śivadvārārgalam): the śivadvārārgala is the brahmārgala. See 2.11 a and 2.12a where brahmadhāma and śivadhāma are identified with one another, and note 245.

402 “by holding the breath” (kumbhakena) : Ballāla (f. 63r11) takes kumbhakena with bhittvā. This seems unlikely since kumbhaka is not mentioned as necessary for piercing the brahmārgala in paalas 1 and 2 while breath-retention is often invoked as the means of forcing Kualinī upwards (see e.g. 2.35, 2.41cd).

403 On praviśet being taken as having a causative meaning see note 399. I understand its object to be Kualinī. It could perhaps be the tongue but I have decided against understanding it thus for two reasons: firstly, there is no need to use kumbhaka to insert the tongue into the passage above the palate; secondly, if Kualinī were not meant here, there would have been little point in mentioning her in verse 1. Ballāla (f. 63v1–2) also takes Kualinī as the object, explaining that she is to be cooled down after being heated up in the course of her awakening: pūrvayvagninā taptā mtena śītā bhavatīti tātparya.

404 “he becomes a Khecara” (khecaratva prajāyate): as at 1.75, Ballāla (f. 63v5) glosses khecaratvam with devatvam.

405 “cheating death” (vañcana kālamtyo): see 3.43–47 for a description of kālavañcana by means of this technique. I have emended µ’s kālamtyuś ca to kālamtyoś ca to avoid adopting the unlikely plural kālamtyūnām found in the other witnesses. Ballāla (f. 63v9) explains the plural as referring to the omens of death that he is about to describe (on which see note 430): bahuvacana tu vaksyamāārianimittam.

406 “wandering throughout the three worlds” (trailokyabhramaam): Ballāla explains trailokyabhramaam with atarikamārgea guikāvat “by way of the atmosphere like [when] a pill (guikā) [is consumed]” (f. 63v9–10). A marginal note describes the pill: sā ca siddhapāradāder vihitā “it is made from fixed mercury etc.”. Cf. the khecarīsiddhi described in MKSK paala 6 which is achieved by means of, among other techniques, a guikā.

407 On these eight powers see note 379.

408 “luminous” (jyotirūpiī): this is an aiśa sandhi form, avoiding the correct jyotīrūpiī which would be unmetrical. Ballāla (f. 65r8) identifies Suumā with the Sarasvatī ī: suumākhyā sarasvatīti. This is very unusual—they are normally differentiated: see e.g. DU 4.7 where Sarasvatī is said to be at the side of Suumā.

409 “free of the qualities of colour and shape” (vararūpaguais tyaktam) : Ballāla (f. 65r9) understands vara, rūpa and gua to refer to consonants, colours and the three guas: varā kakārādaya rūpa śuklādi / guā satvādaya. Surprisingly, for tyaktam, µ has sākam and G has yuktam, both meaning “with”.

410 Cf. HP 3.106, ŚS 5.170. DASGUPTA (1976:97) quotes a passage from ombīpāda (song No. 14) in which “the boat is steered through the middle of the Ganges and the Jumna”. See note 259.

411 “an immortal body forever” (sadāmtatanu): the original reading here may well have been that of G and M, parāmtatanu, altered in most of the KhV witnesses to suggest the idea of liberation in an eternal body. Cf. 3.31d.

412 “the place beyond the Supreme Lord” (parameśāt para padam): i.e. the liga in the vessel of amta described at 2.60a–63b (which is above Parameśa, the Supreme Lord, whose location is taught at 2.57).

413 Only µ reads sudhayā śiśirasnigdhaśītayā; all the other witnesses have variants on atha sā śaśiraśmisthā śītalā “then, she, cool, sitting on a moonbeam”. I might have adopted the picturesque latter reading were it not for siñcantī at 13 a which requires a main verb before the sentence can be ended by a conjunction such as atha (the obvious emendation siñcati is unmetrical).

Ballāla (f. 65v8) interprets siñcantī with dvisaptatisahasranāīgaam amtenāhlādayati, “she refreshes the 72,000 īs with amta”. Kualinī’s return to the mūlādhāra is described at CN 53.

414 “into the Three-peaked Mountain” (trikūe): on trikūa, “the Three-peaked Mountain”, see note 259.

415 “free from the process of time” (kālakramavinirmuktam): Alexis SANDERSON made the emendation kālakrama° which I have adopted here. I have adopted the same form at 3.20b and 20c where only J4 has °krama° and have emended 3.21a like-wise. It is time that is under discussion here, not action, so °krama° is preferable to °karma°. Ballāla (f. 67r3) takes kālakarma to mean time and action, glossing karma with kriyā calanādi.

416 i.e. the yogin should hold his breath to stop it flowing in Iā and Pigalā. He thereby forces it into Suumā (“the place where day and night are suppressed”). Cf. MaSa 44.23cd īdvaya divārātri suumā kālavarjitā, “the two channels are day and night; Suumā is timeless”. See also Dādu sākhī 16.22 (CALLE'WAERT & DE BEECK 1991:174). KhV 3.19 is found at HP 4.42. Brahmānanda, in his Jyotsnā commentary on the verse, understands liga to mean ātman. The next verse (HP 4.43) equates khecarīmudrā with the flow of the breath in the central channel: savyadakianāistho madhye carati māruta / tihate khecarī mudrā tasmin sthāne na saśaya //.

417 This is one of the few instances where I have adopted a reading of βγ (cedam) over that of µα (devam/ligam; J6 has veda while G's (ahorātram avi)cchedam is probably a scribal emendation). S does the same—devam/ligam makes no sense in the light of verse 17—and glosses idam with pratyaka viśva deha vā, “the perceptible universe or the body” (f. 7Ov2).

418 “death is defeated” (kālamtyujayo bhavet) : Ballāla (f. 70v7–8) gives two possible ways of analysing the compound kālamtyujaya : kālasya mtyoś ca jaya kālādhīno mtyur iti vā / tasya jaya, “defeat of time and death, or defeat of that death which is dependent on time”. I have understood it to have the latter meaning.

419 “with the flower of thought” (bhāvapupena) : KJN 3.24–27 lists eight pupas with which the internal liga is to be worshipped. The bhāvapupa is the fourth. The Bhatkālottara (NAK 1-89/NGMPP B 24/59) contains an aapupikāpaala (f. 136v1– f. 137v2) which describes four varieties of this internal and abbreviated Śaiva worship. Only the first includes the bhāvapupa which is last in the list of eight “flowers”. See also Haracarita p.35 ll.5–8 and p.175. (I am grateful to Alexis SANDERSON for providing me with references from these last two sources.) See also KT 59.28–32.

420 “pointing towards the upper mouth” (ūrdhvavaktragām): it may be that µ,’s ūrdhvavaktrakām was the original reading, with the sense of “having the mouth pointing upwards” where “mouth” refers to the tip of the tongue, with which amta is tasted. The reading ūrdhvavaktrakām suggests a plan of the body found in many hahayogic texts (but not in the KhV) in which the amta tasted by the tongue flows through the śakhinī nāī and emerges at the daśamadvāra which is situated at the rājadanta (the uvula). See e.g. p. 11 l.1, where the aperture is called both mukharandhra and śakhinīmukha, and SSP 2.6. Cf. KhV 2.81.

421 “with a whistling sound” (śītkārea): in the sītkārī prāāyāma described at HP 2.54–56 a whistling sound is made as the yogin inhales through his mouth.

422 “in the supportless space” (nirālambe pade): i.e. practising dhyāna without an object. At HP 4.4, nirālamba is mentioned in a list of synonyms of samādhi.

423 “natural yoga” (sahaja yogam): like nirālamba and unmanī (on which, see note 324) sahaja is given as a synonym of samādhi at HP 4.4. In tantric texts sahaja yoga is a state that arises naturally, without being forced (personal communication from Alexis SANDERSON). DIMOCK, analysing the Caryāpādas of the Vaiava Sahajīyas, writes “The state of sahaja is one of utter harmony, in which there is no motion, no passion, and no differentiation” (1991:42 n.3). Ballāla (f.71r4) understands sahaja yogam to mean yoga using the five sahajās described at 2.120–124.

424 “on the circle of sixteen vowels” (oaśasvaramaale): this is the viśuddhacakra at the throat. See e.g. KMT 11.44a–99b, CN 28, ŚS 5.121. Ballāla (f. 100r5–6) writes: atha kahe viśeea śuddhir yebhyas te viśuddhayo ’kārādivisargā oaśārā(oaśārā°] em.; oaśā*a° S)bhidhā svarā cadrakalās teā tatra sthitākhyā yata iti tadākhya /. Placing the chin on the throat is part of the jālandharabandha technique described in note 267.

425 During my fieldwork many people told me that khecarīmudrā was used by haha-yogins to enable them to stay in a state of extended samādhi. Ballāla, commenting on this verse, writes (f. 72v7): ea cirakālasamādhyupāya, “this is the means to long-term samādhi”. He goes on to say that it should be done in a mountain cave, in the ground or in a maha of certain specifications. There should be a śisyasarakaa-grāmakam, “a group of pupils to protect him” (or perhaps “a small village [nearby] to look after a pupil”), yato dehasarakaam āvaśyakam, “because the body [of the yogin] must be looked after [by one of his pupils]” (f. 72v8).

Writing in 1342 CE, Ibn Battūta reported of the jokīs (yogīs): “These people work wonders. For instance one of them remains for months without food and drink; many of them dig a pit under the earth which is closed over them leaving therein no opening except one through which the air might enter. There one remains for months and I have heard that some jogis hold out in this manner for a year” (HUSAIN 1953:164). HONIGBERGER (1852:127–131) recounts the celebrated story of the “faqueer” Hari Dās who in 1837 was buried for forty days in a locked chest in a garden in Lahore. He was exhumed in front of “a great number of the authorities of [Maharaja Ranjit Singh's] durbar, with General Ventura, and several English-men from the vicinity” and revived. Describing those who practise this technique, HONIGBERGER continues (ibid.: 129): “those who do succeed must undergo a long and continual practice of preparatory measures. I was informed that such people have their fraenulum linguae cut and entirely loosened, and that they get their tongue prominent, drawing and lengthening it by means of rubbing it with butter mixed with some pellitory of Spain, in order that they may be able to lay back the tongue at the time they are about to stop respiration, so as to cover the orifice of the hinder part of the fosses nasales, and thus… keep the air shut up in the body and head”. Sir Claude Wade witnessed the revival and reported that Haridas's servant “after great exertion opened his mouth by inserting the point of a knife between his teeth, and, while holding his jaws open with his left hand, drew the tongue forward with his right,—in the course of which the tongue flew back several times to its curved position upwards, in which it had originally been, so as to close the gullet” (BRAID 1850:13). BOILEAU (1837:41–44) describes a similar spectacle that took place at Jaisalmer adding that “the individual… is, moreover said to have acquired the power of shutting his mouth, and at the same time stopping the interior opening of the nostrils with his tongue” (ibid.:43). Cf. the seventeenth-century account given by TAVERNIER (1925:156). MONIER-WILLIAMS (1878:50–53) reports two such attempts at “Samādh”, both duplicitous. In the first, the practitioner's “friends were detected by the villagers in pouring milk down a hollow bamboo which had been arranged to supply the buried man with air and food. The bamboo was removed, and the interred man was found dead when his friends opened the grave shortly afterwards” (ibid.:50). BRUNTON (1995:112–120) describes in detail a meeting with an Egyptian fakīr who used the khecarīmudrā technique to enter a state of catalepsy. The technique, says the fakīr, was originally developed by Indian yogins. At every Kumbh Melā since the 1992 Ujjain Sihasth, a yogin called Pilot Bābā, together with a Japanese disciple, have remained in an open pit for periods of up to a week, emerging with much ceremony in front of large crowds. See also SIEGEL 1991:168–170.

426 “at that place” (tatpadam): Ballāla f. 72r1–2 understands tat to be referring to brahman: o tat sad iti trividho brahmanirdeśa.

427 “lifeless” (nirjīvavat): cf. AY 1.39, KJN 14.82–85, MVUT 17.22c–23b, Svāyabhuvasūtrasagraha 20.33–35 (as edited by VASUDEVA, 2004:43 5). Ballāla (f. 71v10) notes the objection that if the body seems lifeless then surely a bad smell and other signs of putrefaction (daurgandhyādi) that are found in a corpse will arise. But this is not the case, he says: it is contraindicated by the use of bhāti (his reading for bhāvi°).

428 Ballāla (f. 72v6–9) describes the yogin's state here as samādhi and mentions in passing some bizarre techniques for both reaching and returning from samādhi practised by other schools: anye bahvabhyāsena jñātābhyataranāīviśeamardanenāpi ta kurvati kārayati ca / eke tu śavāsanasthit* ā sv* obhayapādāguhāgrātarmanasai gratāyā ca ta kurvati /… tatra samādhyavatāropāya bāhyavāyusparśa śirasi navanītaghtādimardana / tadavatāravelāyā devamūrttyādi tannetrāgre dhārayen na śiyādis tihed… /. “Others, after lots of practice, use a special massage of an internal channel that they have discovered to enter [samādhi] (and cause others to enter it). Some enter it in the corpse pose, once they have focussed their minds on both their big toes… The touch of fresh air [or] massaging the head with butter, ghee etc. are the means of bringing [the yogin] round from samādhi. When bringing him round one should hold an image of a deity or such like in front of his eyes. Pupils etc. should not stand [in front of him].” At the 1998 Hardwar Kumbh Melā, Raghuvar Dās Jī Yogīrāj tried to induce samādhi in me by squeezing the sides of my neck. I backed away as I started to feel faint. The corpse pose practice is taught at DYŚ 46–48.

429 “which consists of an eternal body” (nityadehamayam) : Sαβγ have nitya° her where µ and G have tyaktvā. This indicates a doctrinal difference between the earlier and later recensions of the text. In µ and G śivatvam happens after death; the later tradition wants śivatvam in an eternal body. The original idea behind 31 ab was of Kualinī breaking out of the top of the skull (resulting in physical death for the yogin) rather than just entering the abode of Brahmā (as has already been described at 28cd). The use of vrajati (in contrast to vasati at 28d) confirms that this was the meaning intended in µ and G.

The readings for 31d found in the later tradition are slightly awkward. That of α2, nityadehamayam, is better than the nityadeham imam of the other witnesses and I have thus adopted it.

430 Cf DYŚ 251–258 (=YTU 107-111), which describes how the yogin can leave and return to his body at will.

Ballāla here embarks on a long excursus about ariāni, ways of forecasting impending death, including, among several others, palmistry, pulse-reading, dream analysis and shadow-inspection (f. 73r1–f. 75v4). (Ariajñāna (or kālajñāna) is also taught at MVUT 16.48–52, Dīkottara samudāyaprakaraa 59 (VASUDEVA 2004:361), Matagapārameśvara yogapāda 4.98cd–100ab, 4.127–144, KMT 23.1–80, Mārkaeyapurāa adhyāya 43 (of which vv. 3–26 are cited at ŚP 4564–4590), VS adhyāyas 7–8, YBD 11.135–143, in the ninth upadeśa of the ten-chapter HP (H3) and in vv. 761–772 of the Jogpradīpakā. On aria in medical literature see MEULENBELD 1974:442.) Then, at f. 75v4–7, Ballāla gives three ways (corresponding to the techniques described in KhV 3.32c–43b, 43c–47d and 48a–55b) in which the yogin might deal with impending death: ittha kālamtyum upasthita vijñāya yadā yogino buddhir ima deha tyaktu bhavet sā ca trividhā / samādhāv aikyabhāvinī kālavacanī atyatamokagāminī ca / tatrāpi prathamā dvidhā / svadehe jīveśaikyabhāvinī / paradehe svātmabhāvinī / aichikā parakāyapraveśarūpā ceti// dvitīyāpi dvidhā / kevalasamādhyā kālātikramātā / sarvadhāraayā tadatā ceti / ttīyā tu svechotkrātyā brahmaikyasapādinī /. “Having thus realised that death is at hand is ‘when the resolve of the yogin is to abandon this body’. And it [i.e. the resolve] is of three kinds: bringing about unity in samādhi, cheating death and going to final liberation. Of these, the first can take two forms: bringing about in one's body the union of the vital principle with the Lord, and manifesting oneself in the body of another (this takes the form of willful entry into another's body). The second can also take two forms: by means of the highest samādhi until the time [of death] has passed, and by introspection of all [objects] until that [time] has passed. The third brings about union with Brahmā by means of voluntary yogic suicide.”

431 I have taken jīvānilam as a dvandva; Ballāla (f. 76v3) takes it as a karmadhāraya (see also note 251).

432 Ballāla (f. 76v4–7) adds that the yogin is to inhale with the right nostril: pigalāmārgea pūrakapūrvaka kumbhaka saprāpya.

433 The Svādhihāna centre is in the region of the genitals—see e.g. GŚN 22. Here Śiva is describing Kualinī’s ascent through the six centres that are the basis of a system of subtle physiology found in some hahayogic texts (e.g. GŚN 15–16, ŚS 5.77–155, Yuktabhavadeva 3.234–252—a seventh centre, the Sahasrāra, is added in these texts) and which has become today the most widely accepted model of the body. In the texts of hahayoga there are many different systems of cakras (see e.g. SSP 2.1–9 which lists nine cakras and KAVIRĀJ 1987, who describes a list of 26 cakras given in a manuscript of the Vairāapurāa) and this reflects the even greater variety of such systems found in earlier tantric works. The first systematic description of the six cakras can be found in paalas 11–13 of the Kubjikāmatatantra. (An earlier, but vague, reference to six cakras can be found at Mālatīmādhava 5.2; a description of raising the breath through six centres which are not called cakras is given at Bhāgavatapurāna 2.2.19c–21d). WHITE (1996:134) suggests that the earliest systematic description of the six cakras is found at KJN 17.2b– 4a. 17.2c–4b reads: ha guhya sanābhiñ ca hdi padmam adhomukham //2/ / samīrastobhaka cakra ghaikāgranthiśītalam / nāsāgra dvādaśānta ca bhruvor madhye vyavasthitam //3// lalāa brahmarandhra ca śikharastha sutejasam /. The text is obscure and possibly corrupt but there are clearly at least eight cakras listed and probably as many as eleven. 17.4cd implies that they number eleven: ekādaśavidha prokta vijñāna dehamadhyata. (In a later article (2003:147) WHITE acknowledges that this passage describes eleven cakras.) As WHITE also notes (ibid.:423 n.86), at KJN 5.25–28 there is another of the text's many lists of centres in the body, which, although again rather obscure, does describe seven centres, of which five have locations similar to those of the cakras in the KhV and other hahayogic texts. At KJN 10.6–8 there is a list of eight cakras of which six correspond to those described here in the Khecarīvidyā.

434 The Maipūra cakra is at the navel. See e.g. GŚN 23. (But cf GŚN 25 where the kanda, which in verse 23 is situated at the nābhimaala and is the site of the maipūrakacakra, is said to be below the navel. This discrepancy (or, perhaps, textual corruption) is noted by Ballāla at f. 99v7.)

435 The Anāhata centre is located at the heart. See e.g. ŚS 5.113.

436 i.e. the Viśuddhi/Viśuddha cakra at the throat. See e.g. KMT 11.44a–99b, CN 28–29. The mixing of metaphors in the description of this “lotus” as “sixteen-spoked” is curious. As a lotus, this centre is usually said to have sixteen petals (GŚN 15c–16b); one would expect it to be called a wheel (cakra) when described as sixteen-spoked. The emendation of padme to cakre is tempting but nowhere else in the KhV is cakra used in this sense.

437 The Ājñā centre is located between the eyebrows. See e.g. ŚS 5.128.

438 “rest” (viśrāmam) : see JRY 4.2.159 and NT 7.13 for similar descriptions of relaxation in the ocean of amta.

439 Here Ballāla has an excursus on parakāyapraveśa, “entering another's body” (f. 77v8– f 79r7). Jogpradīpakā 797–804 describes this practice and Alexis SANDERSON has provided me with the following references to Śaiva passages on this topic: Niśvāsamūla (NAK 1-277/NGMPP A41/14) 7.20; Svacchandatantra 7.328c–329b; Picumata (NAK 3-370) 3.228–232b, 5, 96.19–35; MVUT 21.9–19; 28.294–300; JRY f 195v (vv. 197c–204b), 3.5.31–32b, 4.2.397c–400b; Ligapurāa 1.24.128–130; Vāyupurāa 1.23.209–211. RAMA (1978:437–463) tells of witnessing yogins abandoning their bodies and entering those of others.

440 Kālavañcana, “deceiving death”, is a common motif of tantric and hahayogic texts. Indeed, mastery over death is the sine qua non of the perfected hahayogin: yoga is said to be kālasya vañcanam at GŚN 5–6; the mahāsiddhas listed at HP 1.5– 9 are said to have broken the rod of death (khaayitvā kā ladadam); vernacular tales of Yama's rough treatment at the hands of the Nāths are common (see ELIADE 1969:313–317). Techniques of kālavañcana similar to that of the KhV but using visualisations of amta alone and not involving the tongue can be found at SYM paala 11, MVUT 16.53–54, Svacchandatantra 7.217d–226b, NT 7.37–53 and VS 4.41–46, 6.32–41. The methods taught at KJN 6.16– 28, ŚP 4598–4612, YBD 11.144–161, in the last upadeśa of the ten-chapter HP (H2) and at Jogpradīpakā 773–9 do employ the tongue. At GBS 219ab the tongue is associated with kālavañcana: jibhyā indrī ekai nāl jo rākhai so bacai kāl, “the tongue and the penis [are joined by] one channel; who knows this deceives death”. On the corporealisation of subtle tantric practices see page 27 of the introduction.

441 “knowing the apportionment of [the locations of] death” (kālavibhāgavit): i.e. knowing the division described at 3.44c–45b. Alternatively, the compound could be understood to mean “knowing the apportionment of the time of death”, i.e. having ariajñāna—see note 430. As Ballāla says at f. 79v1, kāla can of course mean both time and death: kālo dvividha yamo ’titādivyavahārahetuś ca /.

442 “death” (kālam) : I am taking kālam to be an aiśa neuter and the subject of vrajati.

443 i.e. with Kualinī in union with Śiva in the ocean of amta above the gateway of Brahmā.

444 “supremely content” (paramasatua) : Ballāla (f 81r8) glosses paramasatua with na tu kicidicchayāsatua / punarjanmaprasagāt / “not unsatisfied because of the slightest desire, because it would [then] undesirably follow that he would be reborn”.

445 “the rock of Brahmā” (brahmaśilām) : this rock (śilā) is perhaps the same as that at the top of the forehead described at 2.25. Ballāla (f. 81r10–11) says that it is like a rock blocking the way to brahman: brahmanirodhakām (em.; brahmao rodhekām Spc, brahmaa rodhakā Sac) śilām iva, and locates it at the crown of the head where the fontanelle is found in infants and where [dead] renouncers’ skulls are to be smashed with a conch shell: yatra bālaśirasi mdula tatraiva ca sanyā sinā śakhena mūrdhā bhettavyo ’tarāla iti. Witness R2 has the variant reading brahmasabhā, “Brahmā’s assembly”.

446 i.e. the yogin is to return the microcosmic elements, mind and sense-organs of his body to their macrocosmic origins. Cf. ŚP 4531–4541.

447 “untouched” (aspa): the conjectural emendation suggested by SANDERSON of ada and its variants to aspa is found in the BKhP as a marginal addition by a later hand (f. 82r2).

448 “the orb of the sun” (sūryasya maalam): this is the only mention of the sūryamaala in the text. Maalabrāhmaopaniad 2.1.5 describes how the agni°, sūrya°, sudhācandra° and akhaabrahmatejo° maalas are seen in the process of śāmbhavī mudrā but these are unlikely to refer to places in the body. Ballāla (f. 82r3) associates the sūryama.ala with the Pingalā ī: sūryamaala pigalā sūryanāī tanmārgea tanmaala pūrakapūrvakakubhakena bhitvā. His interpretation seems forced: sūryasya madalam almost certainly refers to a region at the top of the head. ŚP 4591–4611 describes both videhamukti and kālavañcanā. To deceive death the yogin seals all ten apertures of the body (4602) and floods it with amta. To abandon the body he seals only nine doors (4594) and then, using his breath and his mind, he fires the arrow of his soul by way of the tenth door towards the supreme target (4595–6). This tenth door is in the region of the top of the head (see note 240). Cf. Bhagavadgītā 8.12–13; VS 3.54–56. Descriptions of methods of “yogic suicide” (utkrānti) are found in several tantric Śaiva works. See the testimonia to MVUT 17.25–34 in VASUDEVA 2004 and the editor's analysis on pp. 437–445; in the KMT, the yogin is instructed to perform utkrānti when the place of the uvula dries up (23.99a). Alexis SANDERSON has provided me with the following further references to Śaiva passages on utkrānti: Skandapurāa 182.973–977; Niśvāsakārikā (NAK 1-277/NGMPP A 41/14 f. 114v ff.) paala 33; Sārdhatriśatikālottara (NAK 5-4632/NGMPP B 118/7) 11.13–19b and Rāmakaha ad loc; Bhatkālottara Utkrāntyantyeipaala vv. 1–7 (NAK1-89/NGMPP B 25/49 f. 187v3 ff.); Matagapārameśvarāgama Caryāpāda 9; Picumata (NAK 3-370) patalas 5 and 100; 28.292–302; Jñānasiddhānta (Old-Javanese, ed. and tr. Haryati Soebadio, The Hague, 1971). Yogayājñavalkya adhyāya 10 teaches how to abandon the body by means of samādhi.

449 “absorbed in Śiva” (śive līnah): in the description of utkrānti at Svāyabhuvasūtrasagraha 22.2 d (VASUDEVA 2004:441 n.214) the yogin is said to be śivalīnamanā: his mind is absorbed in Śiva. At ŚP 4596 he becomes absorbed in paramātman.

450 Ballāla (f 82r6–f.82v11) here describes two types of liberation: gradual (kramamukti) and subitist (kevalamukti), citing “Yājñavalkya”, the Tantrarājatantra, the Bhāgavatapurāa, and the Yogasūtra with Vyāsa's Bhāsya.

451 Cf. the description of khecarīmudrā in the Haharatnāvalī where it is said that the yogin abandons his body and enters the place of Brahmā at the end of the kalpa: kāya tyaktvā tu kalpānte brahmasthāna vrajaty asau (f. 13v1).

452 From here to the end of paala three, witnesses µ and G vary considerably from each other and from the text as I have presented it. Analysis of their variant readings indicates that µ preserves the earliest version of the passage and that G represents an intermediate stage between µ and the other witnesses. µ’s passage is in praise of madirā, “alcohol”, and this explicit Kaula ideology has been expunged from the other witnesses who have turned the passage into a eulogy of Khecarī and śivabhakti. See pp. 7–9 of the introduction for a detailed comparison of the different versions.

453 We return here to the oldest layer of the text (see pp. 12–13 of the introduction). Thus in this verse vidyā would originally have meant the mantra of Khecarī but can now be interpreted as meaning the teachings of the whole text.

454 On melanal melaka see note 198.

455 Here Ballāla has an excursus on the various methods of Śaiva worship (f. 85v– f 86(3)v). Among more orthodox practices he includes at f. 86(1)r2–3 a ten-fold physical worship from the Rudrahdaya: kahavikāragadgadākarajihvāspadauhasphuraa śarīrakapanaromācasvedāvalabanānirgamarodana pāravaśyatā. At f. 86(1)r10–f. 86(1)v1 he gives a six-fold mantranyāsa from the Śivārādhanadīpikā to be performed when bathing in ashes: om īŚānāya nama śirasi o tatpuruāya nama mukhe o aghorāya nama hdaye o vāmadevāya nama nābhau osadyojātāya nama pādayo onama sarvāge evam uddhūlayed eva snānaprakrama. At f 86(i)v5 he mentions a pāśupatavrata from the Atharvaśira[upanisad].

456 “to advance in all types of yoga” (sarvayogābhivddhaye): the next nineteen folios of Ballāla's commentary (f 87v–f. 106v) are devoted to a description of sarvayoga, all the various methods of yoga.

457 For descriptions of suitable places for the hahayogin to carry out his practice see e.g. DYŚ 107–114, HP 1.12–13. VASUDEVA (2004:247–251) surveys similar descriptions in Śaiva tantric works. Jogpradīpakā vv. 671–5 says that while training to practise khecarīmudrā, the yogin should stay in a secluded hermitage or forest hut for six months, not speak to anyone, repeat his mantra day and night, eat rice and milk without salt, use the herbal preparations described in its verses 665–670 (which are based on those described in Khecarīvidyā paala 4), cut his frenum every Sunday, perform dohana every fortnight and mathana day and night. His tongue will grow by four fingers, he will attain both bhukti and mukti and have no fear of birth and death.

458 “furnished with all that is necessary for the practice” (sarvasādhanasayukta): Ballāla (f. 108r3) understands sarvasādhana to refer to food and medicinal herbs: svāhārasādhanāni tauladugdhādīni auadhāni śuhyādīni ca, “the requisites for his food [such as] rice and milk etc. and medicinal herbs [such as] dried ginger etc.”. At f. 108r9-10 he says how the yogin is to obtain them: dhanāhyarājāśrayea… svīyadravyea vā, “by recourse to a rich king… or by means of his own wealth”.

459 Ballāla (f. 109r11) expands uvāca with eva karuārdrakaākea tārakopaderā śivena prollāsitā laksyabhinnā pārvatī ta pratyuvāca, “thus gladdened by Śiva, the teacher of salvation, whose sideways glance was wet with [tears of] compassion, Pārvatī, whose purpose had been fulfilled, replied to him”.

460 “whose diadem is the crescent moon” (candrārdhaśekhara) : by this epithet “Śambhu's altruism is proven—when he holds the moon that consists of amta at his heart, there is the destruction of [his] poison and fever, but he holds it at his diadem in order to appease the three-fold afflictions of others.” (amtātmanaś candrasya svahdaye dhāraenāpi viadāhopaśātisabhave sati śikhare dhāraa tu pareā trividhatāpaśātaye eveti lokopakāra siddha : BKhP f. 109v6- 7).

461 “who can be attained [only] by true devotion” (sadbhaktisalabhya): the reading sadbhāva°, “true essence”, found in µW1 and G (after 3.56d) may be original: see KJN 21.10 where, after giving an exposition of the different Kaula schools, Bhairava declares “kathita kaulasadbhāvam”; cf. ibid. 14.93–94 where amta is located in the khecarīcakra and identified with kaulasadbhāva.

462 This chapter is a later edition to the text. See page 12 for details.

The Jogpradīpakā (vv. 665–670) draws on this paala in its description of various herbal preparations useful in the practice of khecarīmudrā.

Four verses in this paala are not in anuubh metre: verse 2 is in vasantatilakā, 3 and 10 are in upajāti, and 4 is in sragdharā. These different metres have in places confused scribes and account for some of the variants and omissions in the witnesses.

463 “the highest limb of the mendicant” (bhikūttamāgaparikalpitanāmadheyam): as explained by Ballāla (f 11OR9–11), this compound is a riddle standing for the muī plant (which is mentioned by name in verses 9 and 12). The mendicant (bhiku) is the sanyāsī whose highest limb (uttamāga), his head, is shaven (mua). He is thus muī. Muī is Sphaerantus indicus Linn. (DASH & KASHYAP 1980:54). (When reporting the botanical names of plants mentioned in this chapter I give only those primarily identified with the Sanskrit term; for alternatives the relevant references in MEULENBELD 1974 or DASH & KASHYAP 1980 must be consulted.) Ballāla (f. 11or9) introduces his commentary on this verse with atha muīkalpam āha, indicating that he regards this practice as a form of kāyakalpa, a technique of physical rejuvenation still practised by hahayogins in which the yogin stays in darkness in a cave or specially built room for long periods (usually a month), restricting his diet to a single herbal preparation. Similarly, his commentary on verse 4 begins atha vārāhīkalpam āha (f. 111r1). Tonics to be consumed in kāyakalpa are described in MS O (on which see p. 54). The second upadeśa of the Yuktabhavadeva contains detailed descriptions of thirteen kalpas, including muī-kalpa. See also the Kākacaīśvarakalpatantra. For a modern account of the technique, see ANANTHA MURTHY 1986.

464 Ballāla (f. 110ov1–2) says that takra here is three parts buttermilk to one part water and cites Amarakośa 2.8.1280 for definitions of the different varieties of takra: takra hy udaśvin mathita pādābv ardhābu nirjalam ity amara /.

465 On the use of takra, āranāla, madhu and śarkarā in Āyurveda see MEULENBELD 1974:465–7, 445, 486–7 and 507–8 respectively.

466 Verses 2 and 4 are written as instructions for a physician attending to the yogin—in this verse the verb is dadyāt, “he should give”, while in verse 4 there is the causative pāyayet, “he should cause to drink”. ANANTHA MURTHY (1986:57–61, 235) explains the necessity of an attendant physician to oversee kāyakalpa.

According to Ballāla (f. 110ov4–5), the yogin should be fed the pills for either 49 or, (“some say”— ke cit), 40 days, in the morning and evening.

467 It seems likely that a half-line is missing at the end of this verse, in which the yogin would have been said to obtain the various benefits listed in pādas 2e and 2f.

468 “vigour” (vīrya): this may refer to semen— MaSa 40.50 describes a herbal rasāyana useful for semen-retention.

469 Ballāla (f. 110v8-9) adds: varāha sūkara sa ca vipraka sūkmam api śabdam avadhārayati, “varāha is a boar, and a boar can make out distant and subtle sounds”.

470 Vārāhī is Tacca aspera Roxb. (MEULENBELD 1974:599–600). According to Ballāla (f. 111r1), vārāhī is known as vilāī in the vernacular (bhāāyām). Under bilāī-kad, MCGREGOR (1995:735) writes “cat's root: a large climbing perennial, Ipomoea digitata, having tuberous roots which are eaten and used medicinally”.

471 “he gets rid of blackness on the body” (kabhedī śarīre): this epithet is odd. Most of the KhV manuscripts read kabhedī śarīram. Śarīram is clearly corrupt—none of the adjectives agree with it, nor can it be taken with a verb. The reading that I have adopted, śarīre, is not much better. The only way I can see to translate it is “on the body” which is quite redundant in the context. The three preceding adjectives must be referring to the yogin (µ’s valipalitaharo for hatavalipalita could perhaps be referring to the therapy but this is very unlikely in the light of kakeśī which must refer to the yogin). For kabhedī, S and α2 read kārśyabhedī, “destroying thinness”, which is probably a scribal emendation of kabhedī. The reading varsabhedī found in µ is perhaps due to a scribal error in which a copyist inadvertently looked back to °varau k earlier in the line, although var could perhaps be understood in its meaning of “seminal effusion” (MONIER-WILLIAMS 1988:926). I have taken kabhedī to refer to the therapy's property of combating kuha, which, as Ballāla notes at f. 111r3, can manifest itself in blackness: kuha śveta ka cety anekavidha /.

472 Guggulu is bdellium, the gum of the Commiphora tree (Commiphora mukul Engl.— MEULENBELD 1974:570).

473 Ballāla (f. 111r9–11) describes the preparation of triphalā in detail: laghveraaphalāny ānīyeat sabharjya kuayitvā tatra vipula jala nikipya pācayitvā vastrātarita ktvā tata uparitana taila saghīyāt tac chuddha taila tena sayukta guggulu māhiākhya tathānya triphalāyuta gadhaka ca triphalā tu

ekā harītakī yojyā dvau yojyau ca vibhītakau /
catvāry āmalakāni syus triphalaiāprakīrtitā

“Get some young castor fruits, parch them a little, grind them, add a large amount of water, cook them, put them in a cloth and take the oil from the top. That is pure oil. The wise [yogin], who knows the qualities [of herbs], should eat the guggulu which is called Māhisa and the other [guggulu], mixed with that oil and triphalā, and sulphur. Triphalā: one harītakī [Terminalia chebula Retz.— MEULENBELD 1974: 610] should be used, two vibhītaka [Terminalia bellerica Roxb.—ibid. 1974:601] and four āmalaka [Phyllanthus emblica Linn.—ibid. 1974:527]. This is called triphalā.”
Ballāla (f. 111r11) takes jarādāridrya° as a karmadhāraya: “the debility that is old age”.

474 Winathia somnifera Dunal. (DASH & KASHYAP 1980:46).

475 S is the only witness to read viśvasarpikā here but since I can make no sense of the other variants and Ballāla (f. 111r12) confidently asserts that viśvasarpikā is a synonym of mothā I have adopted his reading. (The Hindī word mothā means “a kind of grass, Cyperus rotundus, and its tuberous root”— MCGREGOR 1995:836; DASH & KASHYAP (1980:25) give musta as the Sanskrit name for Cyperus Rotundus Linn.)

The MaSa manuscripts insert the following corrupt passage between 6c and 6d: hastinā saha yudhyate // triphalā puskaro vrāhmī (vrāhmī J6) † nisākotilalasanī † / punarnavā vddhatārāna yayusnehamiśritā // amāsāhārayogena. Thus the result of eating the preparation is the ability to fight with elephants, while to be free of disease and death the yogin must eat for six months a mixture of triphalā, puskara (Iris germanica Linn.— MEULENBELD 1974:570), brāhmī (Bacopa monnieri Pennell— DASH & KASHYAP 1980:53), †nisākotilalasanī†, punarnavā (“hog-weed,” Boerhavia repens Linn.— MEULENBELD 1974:575) and vddhatārā (probably vddhadāraka, Gmelina asiatica Linn. or Rourea santaloides Wight et Arn.—ibid.:6oo) mixed with oil. The phrase na yayu is likely to be a corruption of the name of an ingredient of the medicine.

476 Sassurea lappa C.B. Clarke (DASH & KASHYAP 1980:61).

477 A marginal note in W1 (ghtamadhuśarkarā) and two pādas added after 14b in γ (ājya guo mākika ca vijñeya madhuratrayam) say that madhuratraya is ghee, honey and sugar, as does Ballāla at f. 110v3, where he adds that they should be in equal proportions.

478 The identity of kunai is uncertain. It is perhaps kunāśaka (Alhagi maurorum—MONIER-WLLLIAMS 1988:286).

479 Muikā is presumably a synonym of muī (see note 463).

480 Eclipta prostrata (MONIER-WILLIAMS 1988:765).

481 Vitex negundo (MONIER-WILLIAMS 1988:554).

482 Amala is a synonym of āmalaka: see note 473.

483 Eleocarpus ganitrus Roxb. (MEULENBELD 1974:596).

484 As WHITE (1996:170) remarks in the context of this verse, BERNIER reported in the seventeenth century that “certain Fakires… can prepare mercury in so admirable a manner that a grain or two swallowed every morning must restore a diseased body to vigorous health, and so strengthen the stomach that it may feed with avidity, and digest with ease” (1891:321).

485 “the silk-cotton tree” (śālmali°) : Bombax ceiba Linn. (MEULENBELD 1974:602).

486 See note 227.

487 See note 259.

488 The verse quoted is abhaksya bhakayen nityam apeya pīyate sadā // agamyāgamana nitya sa yogī nātra saśaya iti goraka /.

489 The passage cited is not found in the Lonavla edition of the Hahapradīpikā.

490 Ballāla wrote this text himself: āsanāni tu asmābhir yogaratnakārikāsu svaktāsūktāni (em.; °ūkt** S).

491 These citations are usually introduced with sūtre and bhāye.

492 The Vāyupurāa is quoted from regularly in Ballāla's lengthy excursus on sarvayoga at f. 87v–f. 116v, sometimes with “upamanyu” to indicate the source of the citation but often without attribution.

493 The verse cited is not in Avalon's edition.

494 This passage is about the different tastes of amta.

495 I have located some of the citations introduced with “sivena” in the Śivasa.hitā. I have been unable to find those listed here.

496 Śivapajaramārkadeyastotra is a correction of śivapacaratnamārkaeyastotra.

497 The Dattātreyayogaśāstra, Yogabīja and Śivasahitā are paraphrased at many places in the commentary (especially between f. 87v and f. 108v). These have not been reported.

498 This text has been published as the Dattātreyayogaśāstra. These citations are usually introduced with “datta” or “dattātreya”.