¹ cf. Mahābhārata, I, 203. 18–30.Google Scholar

² This incident is also mentioned by Maheśvara in a conversation with Umā at Mahābhārata XIII, 128, 1–6.Google Scholar

³ cf. Mahābhārata, I, 203. 10–17:Google Scholar

tayor vadham samuddiśya viśvakarmānam āhvayat/

dṛṣṭvā ca viśvakarmāṇaṃ vyādideśa pitāmahah/

sṛjyatām prārthanīyeha pramadeti mahātapāh//

pitāmahaṃ namaskṛtya tadvakyam abhinandya ca/

nirmame yoṣitaṃ divyāṃ cintayitvā prayatnataḥ//

triṣu lokeṣu yat kiṃcid bhūtaṃ sthāvarajaṅgamam/

samānayad darśanīyaṃ tattad yatnāt tatas tataḥ//

koṭiśaś cāpi ratnāni tasyā gātre nyaveśayat/

tāṃ ratnasaṃghātamayīm asṛjad devarūpiṇīm//

sa prayatnena mahatā nirmitā viśvakarmaṇā/

triṣu lokeṣu nārīṇāṃ rūpeṇāpratimābhavat//

na tasyāḥ sūkṣmam apy asti yadgātre rūpasaṃpadā/

na yuktāṃ yatra vā dṛṣṭir na sañjati nirīkṣatām//

sā vigrahavatīva śrī kāntarūpā vapuṣmatī/

jahāra sarvabhūtānāṃ cakṣūṃṣi ca manāṃsi ca//

tilaṃ tilaṃ samānīya ratnānāṃ yad vinirmitā/

tilottamety atas tasyā nāma cakre pitāmahaḥ//

See also XIII, 127.1:

tilottamā nāma purā brahmaṇā yoṣiduttamā/ tilaṃ tilaṃ samuddhṛtya ratnānāṃ nirmitā śubhā//

⁴ All eight of the goddesses of this second circle, with the exception of Vibhūtī in the northeast, are said to stand on a moon.

⁵ ‘L'indication de la “danse” paraît être contredite par l'assertion selon laquelle “toutes les divinités” sont dans l'attitude noble: sarvā devatāḥ sattvaparyaṅkiṇyo…’ (Mailman, 1975: 380, n. 7).Google Scholar Umā, in the north, and Vibhūtī, in the north-east, are also said to be dancing (nṛtyantī) (cf. Niṣpannayogāvālī, 73).Google Scholar

⁶ Mallman, (1975: 380)Google Scholar: ‘elle tient l'encensoir’. According to the dictionary of Monier-Williams (1899: 517) citing the Kāraṇḍavyūhasūtra, dhūpakaṭacchuka is ‘a small spoon with frankincense’. See the many references of Edgerton (1953: 164) ad kaṭacchu, who hesitates between ‘spoon’ and ‘bowl’.

⁷ Mallman, (1975: 64, 380).Google Scholar Her name in Tibetan is thig le mchog ma. See Vira, and Chandra, (1967: XII, pl. 23 and p. 47).Google Scholar

⁸ Sādhanamālā, nos. 264–7.Google Scholar

⁹ Lessing, and Wayman, (1980: 129).Google Scholar The tantra is Tohuko no. 747 and may be found in vol. VIII of the Peking edition.

¹⁰ Bhattacarya, (1930: 353, 356).Google Scholar

¹¹ For bibliographic data on available editions of the Hindu texts, see Goudriaan and Gupta (1981: 119 n. 31).

¹² It is noteworthy that the Hindu version of the tantra contains eight sets of eight divinities, while the Buddhist contains but seven.

¹³ The Sanskrit names are taken from Bhattacarya (1930: 360). The Tibetan designations are found at Bhūtaḍāmaratantra, 281–4–3/4.Google Scholar

¹⁴ The mantras are unexceptionable and are of the form oṃ + seed-syllable + the name of the apsaras. The syllables are respectively in, śrī, śrī, śrī, hūṃ, sa, sru, baṃ. This is curious since the first four might naturally be taken as a part of the names which follow. This state of affairs suggests that the first four divinities might be allocated to the cardinal directions of the compass and the second four to the intermediate directions, although the text makes no mention of this.

¹⁵ Or: ‘one abides in [her] presence’. I have taken sdod as representing ‘prati-īkṣ’, although ‘ava-sthā’ is equally probable.

¹⁶ On the Kakṣapuṭa and Nāgārjuna, cf. Goudriaan, and Gupta, (1981: 117 and n. 25).Google Scholar According to Goudriaan, (1978: 256)Google Scholarsiddha Nāgārjuna ‘lived in or about the seventh century’. The Buddhist tantric Nāgārjuna may be the same as the siddha Nāgārjuna, although the problem is unresolved. For legends, see Dash, (1976: 53–66)Google Scholar, also Wayman, (1977: 89Google Scholar, with references) and especially Naudou, (1968: 75–83).Google Scholar

¹⁷ See also Goudriaan, (1978: 309–33, esp. p. 329).Google Scholar

¹⁸ According to Monier-Williams, (1899: 166)Google Scholar, this is the seed of Wrightia antidysenterica.

¹⁹ The Tibetan:

yaṅ ni yi ge'i thu bo sbyin//

hūṃ ni khrag daṅ 'dra ba la//

mthar ni svāhā yaṅ byas na//

lha daṅ mi rnams gsod par byed//

²⁰ In the kriyātantra class Bhūtaḍāmaratantra in the chapters devoted to the mantras and sādhanas of the group of eight apsaras, Tilottamā is the second and Rambhā the sixth. See Bhūtaḍāmaratantra, 281–3–5/282–2–7.Google Scholar

²¹ On ākarṣaṇa, cf. Goudriaan, (1978: 294–309).Google Scholar

²² ākarṣaṇaṃ/ oṃ hūṃ svāhā.

²³ cf. Snellgrove, (1959: II, 112, n. 6).Google Scholar

²⁴ For printed dattva.

²⁵ daṅ por yi ge'i bdag po bźag//

de rjes mkha' spyod ma de nas//

mthar ni svāhā sbyar byas na//

saṅs rgyas mams kyaṅ dbaṅ du byed//

²⁶ See also the Sanskrit text of I, ii. 6 which provides the seed-syllables of the yoginīs. The fifteenth and sixteenth items aṃ aḥ a re replaced in the Tibetan with a single oṃ!

²⁷ vaśya ākṛṣṭau cotpalaṃ, dbaṅ daṅ dgug la utpala.

²⁸ Not only in the Hevajratantra: see, for example, Candrakīrti, 's Guhyasamājatantrapradīpoddyotanaṭīkā Ṣaṭkoṭivyākhyā (p. 143. 8–8, on chapter 14)Google Scholar: agnimaṇḍalaṃ kaṇṭhapadmastham/tasyodaye raktaraśmi nirgatyārdhayāmaṃ tiṣṭhati/ tadā vaśyākarṣaṇam kuryāt/

²⁹ The first two verses, II, v. 1–2, have been studied by van der Kuijp, (1985)Google Scholar and Eimer, (1992)Google Scholar who have striven to show that they are, even in the Sanskrit text, an accretion of a date later than the composition of the remainder of the chapter. It must also be added that, although they have been successful in this their object, the revised translations of these two verses offered by van der Kuijp—reacting to the renderings of Snellgrove—are not an improvement. Criticizing the admittedly free but not incorrect translation of Snellgrove which van der Kuijp believes ‘stands to be corrected on a few points’, the latter has contrived to provide new translations which utterly disregard Sanskrit grammar. For example: lcd sarvākārasvabhāvātmā maṇḍalaṃ saṃprakāśayet, Snellgrove: ‘[Hevajra,] the substance of all forms, discourses on the maṇḍala’; van der Kuijp, (1985: 85)Google Scholar, ‘Hevajra,… may (he) clarify the maṇḍala (having) the nature of the essence of all forms’, saṃprakāśayet is a 3rd person singular optative and should hardly be rendered by an injunction, -ātmā is in the nominative singular and cannot serve as an adjective to maṇḍalaṃ. True, Edgerton, (1970: I. 101, paragraph 17. 54)Google Scholar furnishes one example where the nominative ātmā is to be construed as an accusative, but the long ā in this example from the Divyāvadāna (403. 30) is metrically required and, in any case, the compound in the Hevajratantra used as an adjective to hevajraḥ makes good sense. That Ṅor chen provides the reading ātma in Tibetan transliteration (van der Kuijp, 1985: 84) can hardly be taken seriously as against the grammatically correct and semantically feasible readings of the three manuscripts used by Snellgrove. In short, in the study of Indian texts, that Tibetan data or Tibetological reasoning may be helpful or contradictory is a subsidiary matter. When dealing with a Sanskrit text, Sanskritic and Indological considerations must be paramount. This is a principle which all too often of late has not been adequately observed by Buddhologists.

³⁰ The Tibetan:

bo la gźib par mdzad nas ni// dam du 'khyud ciṅ 'o mdzad pas//

bcom Idan gus pas mchod nas ni// bdag med ma yis sṅags źus so// 38

bud med mams kyi dbari byed sṅags// de bźin gdug pa bsdigs pa daṅ//

klu rnams bsdigs par byed pa ñid// lha daṅ lha min rnam ñed pa// 39

de ni ṅa yis bśad bya yis// lha mo bde ba sbyin ma ñon//

saṅs rgyas byaṅ chub sems dpa' la// bdag gis gźan du ma bstan no// 40

'di yi sṅags kyis gaṅ byuṅ daṅ// rdo rje sems dpas gaṅ mdzad pa//

lha mo śin tu 'jigs pa ste// ṅo mi tshogs pas khyod la bśad// 41

'bar ba'i phreṅ ba 'khrug pa yis// dkyil 'khor rab tu bźeṅs nas ni//

rdo rje sñiṅ po dbaṅ sbyin phyir// thig le mchog ni bkug pa ñid// 42

bzlas pa khri yis gsal ba daṅ// sgra riṅ ba daṅ bzaṅ ba yis//

kye'i rdo rje sbyor Idan pas// btsun mo thams cad dgug par bya// 43

bzlas pa 'bum gyis rnal 'byor bdag// kye'i rdo rje'i mal 'byor gyis//

dogs pa med pa'i sems kyis ni// 'di yis las mams thams cad byed// 44

rig byed rnams kyi daṅ po sbyin// zla phyed thig les mam par rgyan//

de nas/

oṃ aṣṭā na nā ya piṃga urdha ke śa bartma ne/ 45 tsa tu rbiṃ śa ti ne trā ya/ de'i rjes ṣo ḍa ṣa bhu jā ya/ kṛṣṇa dzī mū ta va pu ṣe ka pā la mā lā ne kaṃ dhā ri ṇe/ ādyān ta krū ra tsittā ya ardhen du daṃ ṣṭri ne/ 46

oṃ māraya mā ra ya/ kā ra ya kā ra ya gardzdza ya gardzdza ya/ tardzdza ya tardzdza/ śo sa ya ṣo śa ya/ sapta sā ga rān/ bandha bandha nā gāṣṭa kaṃ grhna gṛhṇa śatūn/ ha hā hi hī/ hu hū/ he hai ho hau/ haṃ haḥ phaṭ svāhā// 47

³¹ MS. A, dāntaṃ; MS. B, ?dāptuṃ; MS. C, dātuṃ.

³² MSS. A, B, kṛṣṇatilottamā; MS. C, ākṛṣṇā tilottamāṃ.

³³ No closing quotation marks are found in the printed translation.

³⁴ At this point, Snellgrove inserts a footnote in which he states that ‘… the text inserts a śloka, which seems to have nothing to do with the present context.’ This is incorrect. See infra.

³⁵ This compound is only attested in the Mahāvyutpatti, no. 2440 (Edgerton, (1970: 140)).Google Scholar

³⁶ Quoted by Edgerton, (1970: 140).Google Scholar

³⁷ Monier-Williams, (1899: 205)Google Scholar. The Petersburg Dictionary (v, 1192) adduces Kathāsaritsāgara, 54.173 and defines uparodha as ‘Rücksicht’. This Edgerton, (1970: 140)Google Scholar deems ‘inaccurately defined’. Nevertheless, the Kathāsaritsāgara passage does seem to allow for, indeed, demand, such a translation for uparodha. The context is as follows: the goddess who lives in the Vindhya mountains (vindhyavāsinī) has (at 54. 163) given Yasovarman a boon to consist either of wealth (artha) or enjoyment (bhoga). In order to decide which to accept, he first goes to the house of the merchant Arthavarman where he is given various delicacies to eat, of which the host, evidently for reasons of politesse, also partakes. Nevertheless, the latter states that what he himself has eaten is beyond what he should eat, having a bad digestion, but that he has done so ‘out of regard for you’ (adya tvaduparodhena samāṃsavyañjanaṃ mayā/ bhuktaṃ stokaṃ ghṛtasyārdhapalaṃ bhuktaṃ ca saktavaḥ// sadā tu ghṛtakarṣaṃ ca saktūṃś cāśnami kevalān/ ato 'dhikaṃ me mandāgner udare naiva Jīryate//).

³⁸ Snellgrove, (1959: 1, 112, n. 2).Google Scholar

³⁹ See supra, n. 7.

⁴⁰ Mailman, (1975: 272)Google Scholar: ‘Elle est toujours bleue ou noire…’

⁴¹ cf. II, iii. 60c: vajraṃ bolakaṃ khyātaṃ.

⁴² cf. Guhyasiddhi, 7. 8:Google Scholar

yadi vātha na śaknoti ākṛṣta [°kraṣṭuṃ] divyayoṣitāṃ [°taḥ]/

pīḍyate caiva kāmena tadā seveta mānuṣīṃ//

lha rdzas btsun mo 'gugs pa yi// nus pa gal te med gyur na//

'dod pas gnod pa ñid gyur la// des na mi yi bu mo bsten//

⁴³ Lessing, and Wayman, (1980: 37)Google Scholar have the Buddhas give the consecration to the Bhagavat. However, this is grammatically open and contextually somewhat less likely, since usually it is a single officiant who initiates the disciple. Moreover, taking dṅos as ‘concretely’, as they in fact do and as its position in the sentence supports, it is hard to see how the Buddhas could do this if the consecration involves conjoint action with Tilottamā. See also op. cit., p. 38, 11. 14–16: saṅs rgyas thams cad kyis lha'i bu mo thig le mchog ma bkug nas dbaṅ gsum pa śes rab ye ses kyi dbaṅ dṅos dbaṅ du bskur ba'i tshul bstan/. Note that here the method of the concrete consecration is taught by the Buddhas. The text does not say that the Buddhas give the consecration itself. In a note these authors also refer to mentions of Tilottamā by the Kloṅ rdol bla ma (Collected works, ma, f. 4b–7/5a–1) and in the lexicographic work Abhidhānaśāstraviśvalocana (Derge, ed., vol. po, 78b–4/79a–4)Google Scholar. See also the list of apsaras at Mahābhārata, I, 59. 47–9.Google Scholar

⁴⁴ MS. A, vadantī tatra yoginī; B, vadanti yoginyas tatra.

⁴⁵ MS. A, adds sādhu.

⁴⁶ MSS. A, B, tatrāmilitavyaṃ.

⁴⁷ Leaving out yadi and kathayanti, one has a possible half-verse.

⁴⁸ Printed mālā (m a) bhipreṣitām. All manuscripts as above.

⁴⁹ MS. C, tiṣṭhati.

⁵⁰ Printed bhajeti; A, bahyati; B, bahye: C, bāhyeti; Tib. phyi rol źes, which lacks a marker for the feminine, suggests bāhya-iti. However, note that such is also omitted in brtul źugs bzaṅ for suvratā. In any event, bhaja is unlikely.

⁵¹ Printed āśritya; A, āśsritā; B, āśritāyāṃ; C, āśritāyānaḥ. Note that printed āśritya yields an unwanted long seventh syllable.