Notes

n.1The Sanskrit title of this work on the title page of this translation reflects the emended reading from the Tohoku catalog.

n.2Tarthang Tulku 1982, p. 198.

n.3There are counterparts to many of the forms of Mārīcī described in these texts among the thirty-seven sādhanas for the goddess Mārīcī preserved in the Tengyur that were translated in the twelfth century, and it might be reasonable to assume that Toh 565 and Toh 566 were translated during the same period.

n.4Lewis R. Lancaster, K 1156, The Korean Buddhist Canon, accessed September 5, 2023.

n.5This matter might be settled by a close comparison of the Chinese and Tibetan translations of these works, which we have unfortunately not been able to complete for this publication.

n.6See n.­14 to this translation.

n.7Multiple Sanskrit witnesses are preserved among the Sanskrit manuscript collections at the University of Tokyo, Cambridge University, and the Kathmandu National Archive.

n.8The point at which the Sanskrit cuts off corresponds to Degé Kangyur volume 90, folio 178.a7.

n.9NE 1480/9 reads yāddṛstāpustakaṃ dṛstāstādṛsatvā [sic for yathādṛṣṭaṃ pustakaṃ tathā likhitaṃ?] mayā | yadi suddham asuddham vā mama dokho [sic for doṣo] na dīyate | śrī 3 śrī 3 vajrāccāryyaravṛndrabhadrare [sic for ravīndrabhadreṇa?] lekhyāko yo postakaḥ śubham. This fragment of the scribal colophon might be tentatively translated, “I have copied this down exactly as it appeared in the book. If it is pure or corrupted, the fault should not be placed on me. This was copied down by Śrī (3) srī (3) Vajrācārya Ravīndrabhadra. May there be prosperity and good fortune.” The identification of Ravīndrabhadra as the scribe for this text is traced to the NGMCP catalog card. Given the high honorific prefixes that precede this name in the scribal colophon, it is possible that the text was transcribed for (and not by) the Vajrācārya Ravīndrabhadra.

n.10Mārīcīkalpa, NGMCP E 1480/9, Nepal National Archive, Kathmandu.

n.11Mārīcī­kalpa­tantra, IASWR MBB II 112.

n.12Following D rnam par snang mdzad kyi rgyal po chen pos yongs su gzung ba’o. S and Y read rnam par snang mdzad kyi rgyal po yongs su bzung ba’o, and NE 1480/9 reads vairocanī mahāmudrā samādhāya vicintayet. This translation is tentative, and it is not at all clear to whom this line refers. NE 1480/9 preserves a very different reading that might tentatively be translated, “[o]ne should enter samādhi and visualize the great consort Vairocanī.” While also problematic, the reading in NE 1480-9 makes better sense as the opening line of instructions on visualizing the goddess Mārīcī.

n.13Following S, Y, and K zla ba dang nyi ma dag gi ’od zer las yongs su rdzogs bzhin pa’i lha mo ’od zer can ma. D reads zla ba dang nyi ma gnyis dag gi ’od zer las yongs su rdzogs bzhin pas lha mo ’od zer can ma, and NE 1480/9 reads candrasūryaraśmibhireva mārīcīdevatā niṣpādyamānāṃ vicintanīyā.

n.14D reads dgu phrag gsum yin no, and NE 1480/9 reads tṛnavakāṃ (sic for trinavakāṃ) bhavati. This somewhat cryptic line refers to the group of nine qualities that the Buddha Śākyamuni uses to describe the goddess Mārīcī in The Mārīcī Dhāraṇī (Toh 544, 1.3; Toh 988, 1.3), which he enumerates three times in the text before reciting the first mantra that appears there. It is likely that the current work refers to a work titled The Mārīcī Dhāraṇī that is related to, but does not exactly match, the witness in the Degé Kangyur. This is supported by the fact that the opening narrative to The Mārīcī Dhāraṇī described here is not an exact match for the opening narrative in the version of the witness preserved in the Degé Kangyur, which enumerates eleven qualities.

n.15Following the reading in NE 1480/9. D and S read pa tA kra ma si.

n.16D and S read dkon mchog gsum dang ldan pa zhes bya ba’i, and NE 1480/9 reads triratnarati nāmaḥ. The Sanskrit manuscript clears up a bit of ambiguity in the Tibetan: here it is clear that the Tibetan dkon mchog gsum dang ldan pa has dropped the feminine gender (which would read dkon mchog gsum dang ldan ma) and reads nāma (Tib. zhes bya ba) instead of namaḥ (Tib. phyag ’tshal lo). The Sanskrit therefore makes it clear that the object of this homage is in the feminine case and should be translated “homage to she who delights the Three Jewels.”

n.17D and S read ba rA ha mu khi, and NE 1480/9 reads varāhamukhi. Despite some degree of variation in this term throughout the text, the witnesses appear to favor the spelling varāhamukhi, so this spelling has been adopted as the standard transliteration in this text.

n.18This translation is tentative and follows the reading in D and S: bskul tshig bzhi bcu rtsa bzhi ni/ rgya mtsho gsum ni mtshan brjod pa’i oM dang bcas pa’i snying po dang / nyi shu rtsa gnyis pa dang / yi ge drug pa ni ’jigs pa thams cad las srung bar byed do. NE 1480/9 reads catvāriṁśaccodhāni vyaṣṭakam agraṃ vadati sapraśarava hṛdayadvāviṃśatikaṣadakṣara bhavati, preserving an alternate reading that, while corrupted, might tentatively be translated “one humbly (sic for sampraśrayaṃ?) recites the forty invocation-verses on the first day of the dark half of the month (sic for vyaṣṭakām?) and the twenty-two and six syllable heart mantras.”

n.19Following D and S. D reads ang ga raga sngon po dang stod g.yogs sngon po dang, and S reads ang ga raga sngon po dang / stod g.yogs sngon po dang. NE 1480/9 reads nīlakañcakottarājñī (sic).

n.20Tib. dug; Skt. viṣa. In contexts such as this, when the referent is a ritual substance, it is possible that what is being referred to is feces and not “poison.” Note that the Skt. viṣā can be equivalent to viṣ (“feces”).

n.21Following S ba rA ha mu khi and NE 1480/9 varāhamukhi. D reads bA rA hA mu khi.

n.22Following S, Y, and K skyil mo krung du gnas pas lte ba’i phyogs su, which is supported by NE 1480/9 paryaṅkenaopaviṣṭena nābhideśe. D reads skyil mo krung du gnas pa’i lte ba’i phyogs su.

n.23Following D and S mtshan brgyad brjod na. NE 1480/9 reads nāmāṣtaśatakaṃ paṭhitavyaṃ. The latter provides an alternate reading that may be more accurate and would be translated, “one should recite the name one hundred and eight times.”

n.24D reads an+dard+hAM na ma si, S reads aM tard+ha na ma si, and NE 1480/9 reads anta[r]dhānamasi. This transliteration emends the reading in D to antardhānamasi, which reflects both the correct spelling of the Sanskrit term for “invisibility,” antardhāna, and is supported by the readings in S and NE 1480/9.

n.25Following D and S rig pa ’dzin pa’i sde snod las phyung ba’i rtog pa ni ’di yin no. The reading in NE 1480/9, vidyādharapaṭalād udhṛto paṃ [sic for uddhṛta idaṃ] kalparājaḥ, suggests that the Tibetan rig pa ’dzin pa’i sde dnod translates the phrase vidyādharapaṭala, not vidyādharapiṭaka, but the Sanskrit terms piṭaka and paṭala are synonymous here and can both refer to a “basket.” It is also possible, of course, that the Sanskrit source text for the Tibetan witnesses originally read vidyādharapiṭaka.

n.26D reads bcom ldan ’das ma ’od zer can gyi rtog pa rdzogs pa’o, S reads bcom ldan ’das ma ’od zer can gyi rtog pa rdzogs so, Y reads bcom ldan ’das ma ’od zer can gyi rtog pa rdzogs s+ho, and NE 1480/9 reads mārīcībhagavatyāḥ kalpaḥ samāpta. This is one of a handful of markers found throughout this text that indicate that this ritual manual is either a collection of shorter texts or, perhaps, that section headings such as this are relics from the larger source text for this ritual manual.

n.27D and S read rang gi snying gar zla ba la laM rnam par bsams la, and NE 1480/9 reads svahṛdcandramaṇḍale āṃ[sic for oṃ?]kāraṃ vicintya. The translation “moon disk” for the Tibetan zla ba is informed by the reading candramaṇḍala in NE 1480/9. It is also worth noting that the seed syllable in NE 1480/9 appears to be āṃ, not laṃ.

n.28Following S de va tA yai, which is grammatically correct. D reads mA rI ts+yai de ba dat+tA ye, and NE 1480/9 reads devatāyā.

n.29Following NE 1480/9 bandhāmi, which is grammatically correct. D and S read ban d+ha mi.

n.30Following NE 1480/9 devatāyai, which is grammatically correct. D reads de ba dat+tA ye, and S reads de ba tA ye.

n.31Following S ba rA ha mu khi and NE 1480/9 varāhamukhi. D reads bA rA hA mu khi.

n.32D reads nub mo mdun du zla ba la/ nyin mo ni nyi ma’i dkyil ’khor la sngon du gsungs pa bzhin du rang gi gnas su bya’o, and NE 1480/9 reads rātrau candrasya devātan [sic] suryamaṇḍalasyāgrataḥ pūrvoktaṃ kārya svasthāne. The meaning of this verse is not entirely clear, but it likely refers to instructions for depicting the goddess Mārīcī. Unfortunately, since this text purportedly consists of excerpts from a larger text, it is not exactly clear which instructions the text is referring to here.

n.33D reads saM pa mUr+ti, S reads sarba mUrd+hi, and NE 1480/9 reads sampra[?]muñcaṭi. This transliteration is based on the reading in D, but the readings in the Tibetan witnesses appear to be corrupted. The tentative reading sampramuñcati in NE 1480/9 might be translated “sets [them] free” (Skt. sampramuñcati). Mārīcī herself is most likely the implied subject of this verb, but because it is not clear what the opening terms ālo, kālo, and tālo mean, it is not in fact possible to determine the correct subject and object related to this verb with any degree of certainty.

n.34Following S ba rA ha mu khi and NE 1480/9 varāhamukhi. D reads bA rA hA mu khi.

n.35Following S pa dA kra ma si. D reads pa tA kra ma si, and NE 1480/9 omits.

n.36Following S tsI va ra ma si. D reads tsI va tA ma si, and NE 1480/9 reads vīvaramasi.

n.37Following S ba rA ha mu khi and NE 1480/9 varāhamukhi. D reads bA rA hA mu khi.

n.38Following S pa dA kra ma si. D reads pa tA kra ma si, and NE 1480/9 omits.

n.39D reads tsI ba rA ma si, S reads tsi va rA ma si, and NE 1480/9 omits. This transliteration is emended based on the precedent for spelling this term already set in this text, particularly in S, and based on the conventional Sanskrit spelling of the term cīvara. Unfortunately, the mantra section in NE 1480/9 does not contain the exact same repetitions of Mārīcī’s mantra that are preserved in the Tibetan witnesses, and this particular instance of the term cīvara appears to be omitted.

n.40Following S ma hA cI va ra ma si. D reads ma hA tsI ba rA ma si, and NE 1480/9 omits. Unfortunately, the mantra section in NE 1480/9 does not contain the exact same repetitions of Mārīcī’s mantra that are preserved in the Tibetan witnesses, and this particular instance of the term cīvara appears to be omitted.

n.41Following S ban d+hA mi, which is grammatically correct. D reads ban d+ha mi, and NE 1480/9 reads bandhami.

n.42Following S ba rA ha mu khi, and NE 1480/9 reads varāhamukhi. D reads bA rA hA mu khi.

n.43Following NE 1480/9 granthiṃ bandhami, which is supported in part by the readings in S (gaM d+ha baM d+hA mi), Y, and K (kan d+hi ban d+ha mi), as well as the reading of this phrase that has already appeared in this text. D reads ban d+ha ban d+ha mi. The reading bandhami is emended to the grammatically correct reading bandhāmi.

n.44Both the Tibetan and Sanskrit contain quote markers after each of these short mantras. Rather than include literal translations of each of these quote markers, the phrase “any of the following mantras” has been added to the English translation to indicate that any of the individual mantras can be recited for one’s own protection.

n.45Following NE 1480/9 cīvaramasi, which gives the correct spelling. D reads tsI va rA ma si, and S reads tsi va ra ma si.

n.46D and S read mgal me’i ’khor lo la gnas pa lcags kyu dang zhugs pas dgug par bya ba, and NE 1480/9 reads alātacakrāruṭaṃ pāśāṅkuśenākarṣayanti vicintyā nayati. This translation is tentative and interprets the object of the phrase “standing in a whirling firebrand” to be the target of the rite. It is also possible to interpret the object of this phrase as the mantra that is in one’s own heart.

n.47D reads tsan+dana dkar pos yan lag byugs la/ me tog dang, S reads tsan dana dkar pos yan lag byugs la/ me tog dang, and NE 1480/9 reads sitapuṣpacandanaliptā[ṃ?]. This translation is informed by the reading in NE 1480/9, where it is clear that the object of these actions is in the feminine, which means that they are being performed for the statue of Kumārī that is installed in the maṇḍala.

n.48D reads mthe bong la rgya skyegs dang mar nag gis bskus la, S reads mthe bong la rgya skyegs dang mar nag gis bskus pa la, and NE 1480/9 reads aṅguṣṭarajitāla[sic for rañjitāṃ?]raktakena tailañca. The object of this sentence is ambiguous, and it may refer to the thumbs (or big toes) of either the person performing the rite or the goddess Kumārī.

n.49D and S read sngags brgyad brgya bzlas te/ dbang bskur ba bstan par bya’o. NE 1480/9 reads aṣṭaśatajaptena [a]bhiṣiktaṃ da[r]śayati, which confirms the reading “consecrated” (abhiṣiktaṃ) for the Tibetan dbang bskur ba. Since it is declined in the masculine case, it refers to the officiant of the rite.

n.50Following S, Y, and K oM mAH. D reads oM mAMH, and NE 1480/9 reads oṃ māṃ.

n.51D reads sngon bzhin du mthong bar ’gyur ro, S reads sngon bzhin du mngon par ’gyur ro, and NE 1480/9 reads purvavan paśyanti, which clarifies that the verb is in the third person plural, letting us know that it is the plural “mātṛs” (mātraḥ) that see the one performing the rite.

n.52Following NE 1480/9 khadirāgnaiḥ tāpayet. D and S read seng ldeng gi me la bdug par bya’o, which reflects a Sanskrit back-translation of *dhūpayet, meaning “should burn incense.” The reading tāpayet, translated here as “should heat up,” makes far more sense in this context.

n.53Following S, Y, and K steng nas skye ba bcings pa’i thod pa, which is supported by NE 1490/8 urdhvahṛta[?]kapāle. D reads steng nas mgul pa bcings pa’i thod pa, which might tentatively be translated, “the skull, which is bound at the neck from the top.”

n.54Following D and S rgyab sprad. NE 1480/9 reads parāmukhī kṛtvā. The Sanskrit parāmukhī literally means “facing away from each other.”

n.55D and S read ’phrog byed dang / dka’ zlog ma dang bcas pas dur khrod du sa brkos la sba’o, and NE 1480/9 reads harasya durgāyā saha śmaśānaṃ nikhāniti. The instructions are a bit unclear here, but it appears that this is another brief rite for sowing discord.

n.56D and S read ’od zer can chen po ’byung ba’i rgyud, and NE 1480/9 reads mahāmārīcyudbhave tantre. NE 1480/9 reads udbhave tantre, not jātatantra as it is rendered in the Tibetan transliteration of the Sanskrit title. Also, both the Sanskrit and Tibetan witnesses render the title of the work here as Mahā­mārīcyudbhave, not Māyā­mārīcyudbhave.

n.57Following D arg+haM. S reads argaM, and NE 1490/8 reads raktaṃ. The reading raktaṃ suggests that this is a blood offering, which is highly unlikely at this stage of the ritual.

n.58D and S read nga rgyal dang ldan pas, and NE 1490/8 reads sāhaṁkāreṇa. The Tibetan and Sanskrit witnesses both read simply “with pride,” but this should most likely be interpreted in terms of the pride associated with being the deity.

n.59Following NE 1480/9 purvoktena vidhānena sarvakarmikakūṇḍeṣu, where “fire pits” is clearly in the plural (sic. kūṇḍeṣu). D and S read sngon du bstan pa’i cho gas las thams cad pa’i thab khung du, suggesting this is a single fire pit (thab khung du) that is suitable for any ritual action. The meaning of this line is not exactly clear. Since this chapter appears to be an excerpt from a larger (and currently unknown) work, we do not have a proper referent for the “aforementioned instructions.”

n.60Following S ba rA ha mu khi and NE 1480/9 varāhamukhi. D reads bA rA hA mu khi.

n.61Following S sarba ri pu s+yai b+h+yaM staM b+ha ya. D reads sarba ri pu s+yai b+h+yaM staM b+hA ya, and NE 1480/9 reads sarvaripusainya[ṃ] stambhaya. The reading sarba ri pu s+yai b+h+yaM in the Tibetan sources is corrupt. The reading for this compound in NE 1480/9, sarvaripusainyaṃ, meaning “the entire enemy army,” is likely the correct one.

n.62Following S ba rA ha mu khi and NE 1480/9 varāhamukhi. D reads bA rA hA mu khi.

n.63D reads sarba ri pu s+yai b+h+yaM staM b+hA ya, S reads sarba ri pu s+yai staM b+ha ya, and NE 1480/9 reads sarvaripusainya[ṃ] jambhaya[?] stambhaya. This transliteration emends the reading stambhāya in D to the reading stambhaya in S and NE 1480/9. The reading sarba ri pu s+yai b+h+yaM in the Tibetan sources is also corrupt. The reading for this compound in NE 1480/9, sarvaripusainyaṃ, meaning “the entire enemy army,” is likely the correct one.

n.64Following S ba rA ha mu khi and NE 1480/9 varāhamukhi. D reads bA rA hA mu khi.

n.65This translation is tentative. D and S read bzung, Y and K read gzung, C reads bsrung, and NE 1480/9 reads grahāya. The verb translated as “write” here literally means “to grasp,” “seize,” or “hold.”

n.66D and S read yang phyis ni lci bas mal nye bar byugs, and NE 1480/9 reads paścād gomayena maṇḍalakam uparipta [sic for upalipta]. The Sanskrit reveals that the Tibetan term mal is likely a scribal error for the Tibetan transliteration of maṇḍala.

n.67Following C msnyems dang bcas pas ’jug par bya’o. D and S read bsnyen pa dang bcas pas ’jug par bya’o, and NE 1480/9 reads praviśya. The reading in the latter simply states that the mantrin “should begin,” and the reading in D is, tentatively, that the mantrin “should perform the rite along with the worshiping rite.”

n.68The painting has been set up facing east, so to sit on the east side of the fire pit facing west means that he should face the painting.

n.69D reads hA bya kA bya vA ha nA ya, S reads ha pya ka pya vA ha nA ya, and NE 1480/9 reads havyakavyavāhānāya. This transliteration emends the reading hA bya kA bya in D to havyakavya, based on the reading in NE 1480/9. This emended reading reflects the correct spelling of this compound describing Agni as the “vehicle” (vāhana) for “oblations to the ancestors” (havyakavya).

n.70D and S read me lha bzang po ston ka’i zla ba’i ’od ’dra ba/ dri ma med pa, and NE 1480/9 reads divya[ṃ] śara[da]ṃ candrāgram iva ni[r]malam amṛtadhārāsravant[am]. This translation is informed by the reading in NE 1480/9, which makes it clear that the Tibetan dri ma med pa modifies Agni, not the amṛta that streams forth from Agni.

n.71This translation is tentative and follows D yang chu bsangs la/ zhi ba la ni. S reads yang chu bsangs la zhi ba la ni, and NE 1480/9 reads punar udake nātyudāpayet[?]. The latter reading might be translated, “during the water offering (punar udake) one should make sure not to offer too much (na atyu[d]dāpayet?).”

n.72This translation is tentative and follows the reading in the Tibetan witnesses: D reads ’bar ba’i me tog lha bzhengs pa lta bu gzhon nu ma ’gro ba la bdud rtsi’i rgyun gyis ’bab bzhin pa’i lha mo bsam la, and S reads ’bar ba’i me tog lha bzhengs pa lta bu gzhon nu ma ’gro ba la bdud rtsi’i rgyun gyis ’bab bzhin pa’i lha mo bsams la. NE 1480/9 reads jvalata[sic for jvalita]pāvaka itvotthitānta[?] mārīcīdevate ca pārāsubahuṇāvayamānā[?] vicintyā. While there are problems with the reading in the Sanskrit witness here as well, the Tibetan reading ’bar ba’i me tog lha, which suggests a Sanskrit back-translation *jvalitapuṣpakadeva, is most likely a corruption of the Sanskrit reading *jvalitapāvaka, or “blazing Pāvaka,” an epithet of the deity Agni. The Tibetan also reads the name of the goddess as gzhon nu ma (Skt. * kumārī ) here instead of mārīcī . If we adopt these emended readings, this line might be translated, “Imagine the goddess Mārīcī emerges and sends forth a stream of amṛta for beings just like brilliant Pāvaka.”

n.73Following S, Y, and K stong phrag brgyad sbyin sreg bya’o, which is supported by the reading in NE 1480/9, aṣṭasahasraṃ juhuyāt. D reads stong phrag brgya sbying sreg bya’o.

n.74Following S, Y, and K ’o thug dang. D reads zho thug dang, and NE 1480/9 reads pāyaṃ. The Tibetan ’o thug is interpreted here as a translation of the Sanskrit *pāyasa, or “an oblation of milk, rice, and sugar,” which is supported in part by the reading in NE 1480/9.

n.75This translation is tentative. D reads yang lha’i khang par bzhugs te, S reads yang lha’i khang par zhugs te, and NE 1480/9 reads puna devatā[gṛ]ha praviśya. It is not entirely clear whether this line refers to a temple or shrine dedicated to Mārīcī, but that is presumably the intended meaning.

n.76This transliteration follows D and S b+hA ba. However, the reading in NE 1480/9, kha kha, is very likely correct.

n.77Both the Tibetan and Sanskrit witnesses include the six-syllable Mārīcī mantra here without relating it to any specific set of ritual instructions. D and S read oM mA rI cyai swA hA, and NE 1480/9 reads oṃ mārīcyai svāhā.

n.78D reads lte ba’i phyogs su bdug par bya’o, S reads te ba’i phyogs su gtug par bya’o, and NE 1480/9 reads nāya [sic for nābhiṃ?].

n.79Following NE 1480/9 pāvakāgnidevatā. The form of Agni in D and S, which read mang sa be da’o zhes bya ba’i me lha, is unclear.

n.80D and S read sgyu ma ’od zer can ’byung ba’i rgyud las phyung ba rtog pa’i rgyal po yang dag pa rnam par snang mdzad kyis gsungs pa ji lta ba bzhin du ji snyed pa rdzogs so, and NE 1480/9 reads māyāmāricyutbhavāt[sic for udbhavāt] tantrād vinirgataḥ kalparājo ’yaṃ samyagorocano bhāvito[sic for samyag vairocanabhāṣito?] yathālabdhaḥ samāpta iti. It is once again not entirely clear to whom the name Vairocana refers here, and this matter may not be adequately settled without access to the tantra that is allegedly the source text for this ritual manual.