Notes

n.1Dharmachakra Translation Committee, trans., Destroyer of the Great Trichiliocosm , Toh 558 (84000: Translating the Words of the Buddha, 2016).

n.2Dharmachakra Translation Committee, trans., The Great Amulet , Toh 561 (84000: Translating the Words of the Buddha, 2023).

n.3Here we follow the most popular Sanskrit rendering of the goddess’ name. The Tibetan translations of the text provide the title Mahāśītavaṇī­sūtra (bsil ba’i tshal chen mo’i mdo), which means The Sūtra of the Great Cool Forest. About this shifting morphology, see Hidas 2017, p. 452, note 20 and Skilling 1992, pp. 141–42.

n.4Dharmachakra Translation Committee, trans., Great Cool Grove , Toh 562 (84000: Translating the Words of the Buddha, 2023).

n.5As above, we have provided the most popular Sanskrit name for the text and the Pañcarakṣā goddess it invokes. The Tibetan translation provides the title Mahā­mantrānudhārī (gsang sngags chen po rjes su ’dzin pa’i mdo), which is essentially synonymous with Mahā­mantrānusāriṇī.

n.6Dharmachakra Translation Committee, trans., Great Upholder of the Secret Mantra , Toh 563 (84000: Translating the Words of the Buddha, 2016).

n.7See Sørensen 2006, p. 90 and Skilling 1992.

n.8Pathak 1989, p. 32. The story comes at the very end of the Bhaiṣajyavastu (Toh 1, chap. 6); see Bhaiṣajyavastu Translation Team, trans., The Chapter on Medicines (84000: Translating the Words of the Buddha, 2021), 11.217–11.234.

n.9Schopen 1989, p. 157.

n.10Hidas 2012, p. 24.

n.11Orzech 2002, p. 58.

n.12Lewis 2000, pp. 119–64.

n.13See Herrmann-Pfandt 2008, p. 181 and Kawagoe 2005, p. 18, respectively.

n.14The colophon reads, “The blessed queen of incantations, the dhāraṇī of The Great Peahen, is complete. It was translated and edited by the Indian preceptors Śīlendrabodhi, Jñānasiddhi, Śākyaprabha, and the chief editor and translator Bandé Yeshé Dé. It was then revised and finalized based on the new lexicon. [The translation] was compared and edited in consultation with Sanskrit manuscripts, and the mantras were aligned with those in the Sanskrit manuscripts” (bcom ldan ’das ma rigs sngags kyi rgyal mo rma bya chen mo’i gzungs rdzogs so/ rgya gar gyi mkhan po shI len+dra bo d+hi dang / dz+nyA sid+d+hi dang / shAkya pra b+ha dang / zhu chen gyi lo tsa ba ban d+he ye shes sdes bsgyur cing zhus te skad gsar bcad kyis kyang bcos nas gtan la phab pa/ rgya dpe la gtugs nas zhus dag byas shing sngags rnams rgya dpe dang mthun par bcos pa’o).

n.15Da jinse kongque wang zhou jing 大金色孔雀王咒經 (Mahāmāyūrīvidyārājñīsūtra), Taishō 986 (CBETA; SAT)

n.16Fo shuo da jinse kongque wang zhou jing 佛說大金色孔雀王咒經 (Mahāmāyūrīvidyārājñīsūtra), Taishō 987 (CBETA; SAT).

n.17Kongque wang zhou jing 孔雀王咒經 (Mahāmāyūrīvidyārājñīsūtra), Taishō 988 (CBETA; SAT).

n.18Kongque wang zhou jing 孔雀王咒經 (Mahāmāyūrīvidyārājñīsūtra), Taishō 984 (CBETA; SAT).

n.19Fo shuo da kongque zhou wang jing 大孔雀呪王經 (Mahāmāyūrīvidyārājñīsūtra), Taishō 985 (CBETA; SAT).

n.20Fomu da kongque ming wang jing 佛母大孔雀明王經 (Mahāmāyūrīvidyārājñīsūtra), Taishō 982 (CBETA; SAT).

n.21Fo shuo da kongque ming wang huaxiang tan chang yi gui 佛說大孔雀明王畫像壇場儀軌, Taishō 983a (CBETA; SAT).

n.22Kongque jing zhenyan deng fan ben 孔雀經真言等梵本唐, Taishō 983b (CBETA; SAT).

n.23The relationship between these translations has been discussed in Sørensen 2006 and Overbey 2016.

n.24The reading of Yamarākṣasī follows the Sanskrit edition. The Degé edition reads “Yama’s sister” (gshin rje’i sring mo), which is perhaps a scribal error for gshin rje’i srin mo.

n.25In the “science of beings” (bhūtavidyā) elucidated in Indic medical literature, grahas are treated as a broad class that is further subdivided into an array of beings, such as are listed here, that are believed to possesses people and trigger physical and mental illnesses.

n.26The Sanskrit term given here is kṛtyākarmaṇa, which is parsed into two terms in the Tibetan, gshed byed dang sgyur ba’i las. The term kṛtyākarmaṇa refers to violent rites that employ a class of beings known as kṛtyās as harmful magical agents.

n.27Skt. durlaṅghita; Tib. bsgom nyes. Judging by the Sanskrit term laṅghita (“overstepped, transgressed”) and its rendering into Tibetan as bsgom, which might be derived from gom (“to step or walk”), the meaning may be connected with a hex whose negative effects are felt if stepped over or on.

n.28These three categories refer to fevers caused by imbalances in one of the three humors‍—vāta (“wind”), pitta (“bile”), and śleṣma/kapha (“phlegm”)‍—in Āyurvedic medicine.

n.29The *Mahā­māyūrī­vidyārājñī­sūtraśatasahasraṭīkā (folios 100.a–100.b) states, “ ‘To draw a boundary’ means that a boundary is drawn around the perimeter, and through its power it is not possible to approach.”

n.30The *Mahā­māyūrī­vidyārājñī­sūtraśatasahasraṭīkā (folio 100.b) states, “‘To ‘bind the earth’ means that the cardinal directions, intermediate directions, and the area above one are sealed, such that no opportunity for harm can be sought from any of one’s sides.”

n.31This translation follows the attested Sanskrit term ardhāvabhedaka. The Tibetan term, gzhogs phyed na ba, could also be interpreted as a translation of pakṣavadha, referring to hemiplegia.

n.32The “major” appendages would include the head, arms, legs, etc. The “minor” appendages include the nose, ears, fingers, and toes.

n.33Sanskrit has tiḍi.

n.34This word is missing from the Sanskrit. C and L accord with the Sanskrit.

n.35Tibetan editions spell this as pānasu (pA na su).

n.36Sanskrit has piśāci.

n.37Sanskrit has orohaṇi.

n.38Sanskrit has tile.

n.39Sanskrit has tile after tili, which is missing in the Tibetan.

n.40Sanskrit has time. H and N accord with the Sanskrit.

n.41Sanskrit has time. H and N accord with the Sanskrit.

n.42Sanskrit has viṣṭhande.

n.43Sanskrit has hulu four times, instead of two as in the Tibetan.

n.44Sanskrit repeats kolu kolu hulu hulu vahulu vahulu.

n.45Tibetan spells this vosa (bo sa).

n.46Sanskrit has domadumbā.

n.47Sanskrit repeats hili ten times, rather than five as in the Degé.

n.48Sanskrit repeats mili ten times, rather than five as in the Degé, and precedes these with oṁ.

n.49Sanskrit repeats tili ten times, rather than five as in the Degé, and precedes these with oṁ.

n.50Sanskrit repeats culu ten times, rather than five as in the Degé, and precedes these with oṁ.

n.51Sanskrit has oṁ preceding this first instance of muhu.

n.52Sanskrit has oṁ preceding this first instance of mulu.

n.53Sanskrit repeats mulu ten times.

n.54Sanskrit has oṁ preceding this first instance of hu.

n.55Sanskrit has oṁ preceding this first instance of vā.

n.56Sanskrit has oṁ preceding this first instance of pā.

n.57Sanskrit has oṁ preceding this first instance of jāla.

n.58This is preceded by haraṇi in the Sanskrit.

n.59Sanskrit has kālaṇi.

n.60Sanskrit has kamali.

n.61Sanskrit has maṇḍitike.

n.62This spelling is according to the Sanskrit, K, Y, H, and N. The Degé has sukusukume (su ku su ku me).

n.63It is not clear which god is being referred to here and in similar statements throughout the text.

n.64In the Sanskrit, each of the following names is rendered in the plural.

n.65This follows the Tib. reading longs spyod ldan. The Sanskrit reads bhāgavān.

n.66The Sanskrit version is missing this repetition of nāga le le le.

n.67The Sanskrit version has thusa thusa instead of thusu thusu.

n.68The Sanskrit version has guru guru instead of gulu gulu.

n.69The Sanskrit version has only one instance of cejini.

n.70The Sanskrit version reads agalu.

n.71This second instance of elā melā is missing from the Sanskrit version.

n.72Instead of tilī melā ili mitte ile tili mitte, the Sanskrit reads ili melā tili melā ili tili melā ili mitte tili mitte ili tili mitte. H and N read tili melā ili mitte ili tili mitte.

n.73Following the Sanskrit version. The Tibetan translation has godohikā.

n.74“Agitated” here reflects the Sanskrit praluḍita, rather than the usual meaning of the Tibetan zhan pa, “feeble/weakened.”

n.75K and Y have six repetitions of hu, H and N have four, and the Sanskrit has six, followed by hulu hulu hulu.

n.76This second instance of cejini (tse dzi ni) is absent in the Sanskrit version.

n.77The Sanskrit version has agalu, K and Y have agayu (a ga yu), and N has agaru (a ga ru).

n.78The Sanskrit version and H, K, Y, and N read elā.

n.79This is followed by tili mitte in the Sanskrit.

n.80This spelling is in accord with the Sanskrit version. Degé reads godehikā (go de hi kA). C, H, K, Y, and N all have do instead of de.

n.81The Sanskrit version reads susuddhe.

n.82The Sanskrit version and H, K, Y, and N read maṅgale.

n.83Śrībhadre is absent in the Sanskrit version.

n.84This spelling accords with the Sanskrit versions as well as H, K, Y, and N. Degé has instead sarvānartha­pravādhani (sar+b+ba a nar+tha pra bA d+ha ni).

n.85This spelling is in accord with the Sanskrit version as well as C, K, and Y. Degé has instead sarvāmaṅgala­sādhani (sar+ba a maM ga la sA d+ha ni).

n.86This is absent in the Sanskrit version as well as C, K, and Y.

n.87The Sanskrit version has adbhute atyabhute instead of acyute adbhute adbhyanabhute. C, K, and Y are all missing acyute.

n.88Vimukte is absent in the Sanskrit version. K and Y have a second mukte (muk te) instead.

n.89This spelling accords with the Sanskrit as well as H and N. Degé reads mośakṣani (mo sha k+Sha ni).

n.90The Sanskrit version and H, K, Y, and N read acyute (a tsyu te).

n.91This is absent in the Sanskrit version as well as K, Y, and N.

n.92The Sanskrit version reads mṛtasañjīvani instead of amṛte amṛtasañjīvani as in the Tibetan.

n.93The Sanskrit version and K, Y, and N instead read suvarṇe. The Sanskrit and H, K, Y, and N end with –varṇe, as opposed to the Degé, which incorrectly reads –varṇa.

n.94This is absent in the Sanskrit version as well as K, Y, and N.

n.95The Sanskrit version instead has brahmajeṣṭe.

n.96Śuci is absent in the Sanskrit version.

n.97The Sanskrit version and C, K, and Y read guci.

n.98The Sanskrit version has mitti. K and Y have midte.

n.99The Sanskrit version has mitti.

n.100The Sanskrit version has mitti.

n.101The Sanskrit version has mile.

n.102Mili tili is absent in K and Y.

n.103Tili is absent in the Sanskrit version.

n.104The Sanskrit version has mitti.

n.105The Sanskrit version has cili.

n.106The Sanskrit version has mili instead of mili mitti.

n.107Vili mili mitti is absent in C, K, and Y.

n.108Vili mili mitti vili mili mili is absent in the Sanskrit version.

n.109The Sanskrit version, K, and Y read sutumbā tumbā.

n.110The Sanskrit version has meḍi. K and Y have meṭi. H and N have maḍi.

n.111The Sanskrit version has prāptamule.

n.112This is according to the Sanskrit version. Degé reads idihānga (I di hA nga).

n.113The Sanskrit version, K, and Y have tila.

n.114The Sanskrit version has naṭti.

n.115The Sanskrit version has keli.

n.116The Sanskrit version adds sudumbe.

n.117The Sanskrit version has damile.

n.118The Sanskrit version has santuvaṭṭe.

n.119The Sanskrit version has khalime.

n.120The Sanskrit version has tumbe sutumbe.

n.121This is in accord with the Sanskrit version, H, K, Y, and N. Degé has anamole.

n.122Kubhaṇṭi is absent in the Sanskrit version, H, and N.

n.123Instead of kili tili misti, the Sanskrit has ili kili misti ili.

n.124Reading the Tibetan rgyal po’i pho brang ’khor as the attested Sanskrit rājakulaº.

n.125The phrases “when among enemies” and “when among those who are hostile” are absent in the Sanskrit version.

n.126The mantra in the Sanskrit version reads cili mili kili mili ketumūle buddhavarṇe vusaraṇe vusaraṇe vudāraṇi vudāraṇi kevaṭṭe kevaṭṭakamūle iti savale tuṃbe tuṃbe priyaṅkare āvartta parivartta.

n.127This is absent in the Sanskrit version, K, Y, and N.

n.128This follows the Sanskrit version, H, and N in reading saśramaṇa[em. śravaṇa]­brāhmaṇa­nikāyām­ prajāyam (Tib. dge sbyong dang / bram ze’i skye dgu dang bcas pa). Degé reads dge sbyong dang bcas pa/ bram ze’i skye dgu dang bcas pa.

n.129Skt. arjaka (Tib. ar+dza ka).

n.130The Sanskrit reads ili mili kili mili kili kiṃ dugdhe mukte sumukte ūḍa nāḍa sunāḍa.

n.131The Sanskrit reads ḍadukā karoḍukā.

n.132The Sanskrit has four instances of mili, rather than two as in the Degé.

n.133The Sanskrit has four instances of pili, rather than two as in the Degé.

n.134The Sanskrit reads culu culu.

n.135Cūlu cūlu is absent in the Sanskrit.

n.136The Sanskrit has four instances of ciḍi, rather than two as in the Degé.

n.137Iṭi viṭi khi khi khi khi is absent in the Sanskrit.

n.138This spelling follows the Sanskrit, H, and N. Degé has harani; K and Y have haraṇi.

n.139Instead of jambhemi stambhemi as in the Degé, the Sanskrit reads ca jambhe prajambhe.

n.140Citre is absent in the Sanskrit, C, and H.

n.141The Sanskrit reads khulu khulu.

n.142The Sanskrit reads dhīre dharya instead of virodhaye.

n.143Muru muru is absent in the Sanskrit.

n.144The Sanskrit has the following mantra: suru suru ke cara cara ke biri biri.

n.145The Sanskrit has the following mantra: ela melā ili melā tili melā tiha duha tilimā timā dumā vimadhu sukumbhā sumbhā tumbā samatumbā āḍe nāḍe tila kuñjanāḍe varṣatu devaḥ tilikisi.

n.146The Sanskrit repeats vudāriṇi.

n.147Ficus religiosa. Skt. aśvattha; Tib. khyad par gnas.

n.148The translation of the preceding lines is informed by the syntax and terminology of the Sanskrit text.

n.149The Sanskrit has the following mantra: ili mili kili mili cili kili voli udumbare sudumoḍe busara busara hu hu karañje karañjamūle iti sanatā kuntari kuntāri nārāyaṇi pārāyaṇi paśyani paśya paśyani kapilavastuni iḍivā iḍivā irivā.

n.150The Sanskrit has the following mantra: kīrttimūle eramūle eraṇḍamūle samantamūle aḍanāḍe kuśanāḍe itte mitte pāru aḍakā maraḍakā ilikiśi godohikā udvandhamābhi bhinne medā.

n.151The Sanskrit has tramaṇi trāmaṇi.

n.152The Sanskrit has the following mantra: suru suru suru suru suru suru suru suru suru suru me svāhā.

n.153The second occurrence of vedyumati is absent in Sanskrit.

n.154The position of the repetitions of ru and cu are reversed in the Sanskrit.

n.155The Sanskrit reads sa.

n.156The Sanskrit has the following mantra: sori sori siri siri mati mati hiri hiri mati pele mati pele piṅgale curu curu hataṃ viṣaṃ bandhumati nihataṃ viṣam bandhumati svāhā.

n.157The Sanskrit has the following mantra, which includes the interspersed translated passages: ele mele kile tile mile śile vāse dumbe dumbe varṣatu devaḥ samantena hili mili tuṃbe tuṃbe aṭṭa vaṭṭa paradu vattā varṣatu devo guḍa guḍantu samantenāḍakovatyāṃ aṇḍe maṇḍe tunde tutunde curke vukke iriḍi miriḍi niriḍi ciriḍi hili hili hulu hulu mili mili tule tatale svāhā.

n.158Sanskrit has the following mantra: bale balkale mātaṅgi caṇḍali puruṣa nici nici nigauri gandhāre caṇḍāli mātaṅgi mālini hili hili āgati gati gauri gandhāri kauṣṭhikā vacari vihāri hili hili kuñje svāhā.

n.159In the following list of yakṣas and their locations many of the names are challenging to interpret in Tibetan translation. We have therefore relied on the names provided in the Sanskrit witness, but it is evident that the Tibetan preserves a number of unique readings. The translations that follows are tentative, and substantial divergences between the Sanskrit and Tibetan have been noted.

n.160This term is rendered as it appears in the Sanskrit text, but the same Tibetan term, ka ba, is used below to translate sthalā .

n.161This follows F and S in reading rgyas pa, which aligns with the attested Skt. vipula . The Degé has rgyal ba.

n.162The Sanskrit reads vibhūṣaṇa, which has been emended based on the Tibetan term ’jigs byed.

n.163Here we have followed the Sanskrit in reading amaraparvata , a location also found in lists of sacred sites (pīṭha) in India. The Tibetan term ’chi med sa zhag suggests the Tibetan translators read amaraparpaṭa, a term which is not attested in other sources.

n.164Here we follow F and S in reading ri chen, which aligns with the attested Sanskrit mahāgiri (“great mountain”). Degé and many other versions of the Tibetan read rin chen (“jewel”).

n.165Meaning in the city of Vidiśa.

n.166Varṇabhaṭa , attested in the Sanskrit, is an unlikely equivalent for the Tibetan ’od ma’i ’gram, the meaning of which is unclear.

n.167The Sanskrit reads vṛhadratha but is emended here based on the Tibetan shing rta che.

n.168This line follows the Sanskrit as the Tibetan is difficult to properly interpret.

n.169The Tibetan ambiguously reads ma rdzogs pa (“imperfect/incomplete”).

n.170The Tibetan reads snyems pa can (S: snyoms pa can), which does not appear to be equivalent to the Sanskrit saṇṭīraka .

n.171The translation of these two lines follows the Sanskrit. The Tibetan, which is challenging to interpret, may reflect a slightly different reading: gnod sbyin chen po bong srung sel/ skyugs pa yi ni ri la gnas.

n.172Rauruka follows the Sanskrit and S (ro ru ka). Degé and most other versions of the Tibetan read ho nu.

n.173The Sanskrit reads kolaśodara, which has been emended here based on the Tibetan bum pa’i lto.

n.174The Tibetan reads rab nyams, which does not appear to be equivalent to the Sanskrit śūnya .

n.175This follows the Tibetan reading ti se gangs. The Sanskrit reads kerala.

n.176The Tibetan reads yang dag byed, which does not appear to be equivalent to the Sanskrit śaṅkālī .

n.177The Sanskrit reads āsaṅga, which has been emended here based on the Tibetan ma chags pa.

n.178The Tibetan reads gso ba’i mtha’, which does not appear to be equivalent to the Sanskrit tarukacchaka .

n.179The translation of the final two lines follows the Sanskrit. In the Tibetan these lines continue in the next verse: “Nandika and Pitānandin, / As well as Vīra, Karahāṭaka, / And Lambodara, live in Kaliṅga” (dga’ byed dang ni pha dga’ dang / dpa’ dang gser gyi lag pa dang / ka ling+ka na lto ’phyang po).

n.180The Tibetan reads nag po che (“Mahākāla”), but since he was already mentioned above, we have followed the Sanskrit here.

n.181The Sanskrit reads vanavāsyām.

n.182The Sanskrit reads dhanavara, which has been emended based on the Tibetan nor gzhan po.

n.183The Sanskrit reads aṅgulipriya, which has been emended based on the Tibetan thal mo dga’.

n.184The Degé reads grong khyer drug, which renders the Sanskrit ṣaṭpura , a term already used in the previous verse.

n.185The Sanskrit reads vaiśalī .

n.186The Tibetan reads dga’ ba po, which does not appear to be equivalent to the Sanskrit caritaka .

n.187The translation of lines three and four follow the Sanskrit syntax and thus ignore the locative particle at the end of line three in the Tibetan.

n.188The Tibetan reads ’gro ba drug pa, which does not appear to be equivalent to the Sanskrit siddhapātra .

n.189The syntax and meaning of this verse in both Sanskrit and Tibetan is unclear, and likely corrupt. The English translation is therefore tentative. See Lévi 1915, p. 46 for a brief discussion of these lines.

n.190The Tibetan reads sha thang ba, which does not appear to be equivalent to the Sanskrit kaṭaṅkaṭa .

n.191This reading follows the Sanskrit. The Tibetan reads ba lang ’dzin dang gzhan brnyas dang / sgo ba gnas can rtag pa dang.

n.192The Tibetan reads mya ngan sa na rmongs byed pa, which could tentatively be translated as “Mohana is in Śokabhūmi.”

n.193The Sanskrit reads campaka, which has been emended based on the Tibetan gtum po.

n.194The Sanskrit reads, “Sindhu, known by the name Pāñcika / Dwells in Sindhu (pāñcika iti namnā tu / vasate sindhu sindhuṣu).

n.195The Sanskrit reads uṣṭrapāda, which has been emended based on the Tibetan mche ba rkang.

n.196The Tibetan reads kapuśa (ka pu sha).

n.197The Tibetan reads tshigs lnga ser po, which does not appear to be equivalent to the Sanskrit pāñcālagaṇḍa .

n.198The translation follows the Sanskrit syntax. D reads pad+ma dkar po’i ’od gsal ba; F, H, K, Y, N, and S read pad+ma dkar po’i ’od gsal ba.

n.199The Tibetan reads brtse ba can, which does not appear to be equivalent to the Sanskrit Śarmila .

n.200The Tibetan reads rab ’jigs pa but should perhaps be emended to rab ’jig pa, which agrees with the Sanskrit.

n.201D reads vavvada (bab+ba da); F reads vavada (ba ba da); H, K, Y, and N read vavapa (ba ba pa); S reads vaccaḍa (ba ts+tsha Da) in agreement with the Sanskrit.

n.202D reads vavvaḍa (bab+ba Da); F reads vavaḍa (ba ba Da); H, K, Y, and N read vavapa (ba ba pa); S reads vaccaḍa (ba ts+tsha Da).

n.203D reads ka pu sha; H and N read ka ba sha; S reads kāpiśi (kA bi shi).

n.204D reads pa lha pa; F reads pa lta pa; S reads ba lha ba.

n.205The Tibetan reads phung po’i bdag po, which does not appear to be equivalent to the Sanskrit rāśina .

n.206The Tibetan reads grong gi lhas, which does not appear to be equivalent to the Sanskrit grāmaghoṣa .

n.207Within the Tibetan transliteration of this vidyā, the phrase amitrān mama (“my enemies”) is translated into Tibetan with bdag gi mi mdza’ ba rnam. The term hana means “kill.”

n.208Within the Tibetan transliteration of this vidyā, the phrase ahitaiṣiṇo mama (“those who wish harm me”) is translated into Tibetan with bdag gi mi phan par ’dod pa.” The term daha means “burn.”

n.209Within the Tibetan transliteration of this vidyā, the phrase pratyārhtikān mama (“those who oppose me”) is translated into Tibetan with bdag la phyir rgol ba rnams.” The term paca means “cook.”

n.210As above, the phrase ahitaiṣiṇo mama is translated into Tibetan. The Sanskrit reads nāśaya ahitaiṣiṇo mama (“destroy those who wish me harm”). The term dhu means “shake.”

n.211The Sanskrit has hu for all ten instances instead of ha in the Degé.

n.212Within the Tibetan transliteration of this vidyā, the phrase śatrūn mama (“my enemies”) is translated into Tibetan with bdag gi dgra bo rnams. The verb for this phrase, naśaya, is given in transliterated Sanskrit but has been translated here for clarity.

n.213As above, the phrase śatrūn mama is translated into Tibetan. And, as above, the verb naśaya is given in transliterated Sanskrit but has been translated here for clarity.

n.214The Sanskrit has cukke.

n.215The Sanskrit has maṅgale.

n.216This is absent from the Sanskrit, F, C, J, K, Y, and S.

n.217This is followed by sūryaprabhe in the Sanskrit.

n.218The Sanskrit has dumbe.

n.219Kovida is supplied from the Sanskrit. The Tibetan reads g.yon can.

n.220The Tibetan repeats “bhūta grahas” (’byung po’i gdon), so we have followed the Sanskrit in reading lepaka here.

n.221As above, this translation follows the Sanskrit in reading kṛtyākarmaṇa as a single term, rather than as two terms as they appear in the Tibetan, gshed byed dang sgyur ba’i las.

n.222Tibetan ’joms pa. There is no equivalent to this term the Sanskrit, which reads unmāda (smyo byed) here.

n.223Referring, presumably, to Śākyamuni.

n.224Piṅgalā follows the Sanskrit. The Tibetan reads spre’u ltar dmar ser.

n.225The Sanskrit has vimale.

n.226This reading follows the Sanskrit and the Tibetan reading sha za khra mo attested in F, K, Y, N, and S. Degé reads sha za phra mo.

n.227This follows the Sanskrit. The Tibetan reads lus las byung ma, which suggests the Tibetan translators reads kāyikā.

n.228This follows the Tibetan ’phyang ma, which is used to translate lambā above. The Sanskrit reads kambojī.

n.229This follows the Sanskrit and S (rig ’dzin ma). Degé reads rig ma, which would translate vidyā alone.

n.230The Sanskrit reads vibhūṣaṇā, which has been emended based on the Tibetan ’jigs byed ma.

n.231The preceding set of names have been rendered as they appear in the Sanskrit. The Tibetan translators called each chung ma, “the wife of” the male deity of the corresponding name, e.g., “the wife of Brahmā,” “the wife of Rudra,” and so forth.

n.232This follows the Sanskrit. The Tibetan reads srin po’i chung ma (“the wife of the rākṣasa”).

n.233Because the names of these rākṣasīs are difficult to interpret in Tibetan translation, this list is largely based on the names as attested in the Sanskrit version. It is not always clear if the Tibetan term corresponds to the Sanskrit, thus the translation of this passage is tentative. Significant variations between the Sanskrit and Tibetan terms have been noted.

n.234Padumā is the Middle Indic form of the Sanskrit padma.

n.235Morikā is the Middle Indic form of mayūrikā. The Tibetan reads dkar mo.

n.236The Sanskrit name Nāḍikā corresponds closely with the Tibetan term attested in S, sbu gu can. Degé reads dbu bu can.

n.237The rākṣasīs Vidurā, Gaurī, and Gandhārī do not have a clear equivalent in the Tibetan translations. Following Piṅgalā (Tib. dmar ser mo) the Tibetan reads srin mo thigs pa ’dzin dang / srin mo gtum mo ma dang / srin mo ri ’dzin ma.

n.238Only in S (za ba ma) does aśānī have a clear equivalent in the Tibetan translation. This rākṣasī is omitted in all other Tibetan sources consulted.

n.239Following “Skandha,” the Tibetan translation includes a rākṣasī not attested in the Sanskrit: srin mo mun pa.

n.240Tib. srin mo mche ba ma dang / srin mo yA ma dang. These names are missing in the Sanskrit as well as in Y, N, and S.

n.241The Tibetan reads srin mo stobs bzang ma, which does not appear to be equivalent to the Sanskrit acalā .

n.242The Sanskrit reads caṇḍā, which has been emended based on the Tib. zla ba.

n.243The Tibetan reads srin mo zhags pa ma, which does not appear to be equivalent to the Sanskrit maṇḍitikā .

n.244The Tibetan reads the term asimuṣaladharā as two separate names, *asidharā (ral gri can) and *muṣaladharā (lag na gtun shing ’dzin).

n.245Sanskrit has ten repetitions of hili.

n.246H, K, and Y have taḍa tavaḍo instead of haḍa havaṭo as in the Degé.

n.247Haḍa havaṭo vakṣe vakṣe vakṣe is absent from the Sanskrit.

n.248Sanskrit adds hala hala here.

n.249The Sanskrit reads curu curu.

n.250The Sanskrit reads brāhmaṇa .

n.251This line is not found in F, Y, N, and S.

n.252This line is not found in C, F, J, K, and Y.

n.253This line is not found in C, F, J, K, and Y.

n.254This follows the Sanskrit, F, and S. Degé reads ar+tha sha ba ri, K and Y read ar tha sha ba ri, and C and J read a tha sha ba ri.

n.255This line is absent in the Sanskrit.

n.256The Tibetan Degé version here calls Suvarṇāvabhāsa a peacock “queen” (rma bya’i rgyal mo). We have followed the Sanskrit reading of “peacock king” (mahāmayūrarājāya) in keeping with the gender of this figure earlier in the text.

n.257Tib. klu’i rgyal po dug med bcas. This nāga king is not attested in the Sanskrit version.

n.258This spelling follows the Sanskrit. Degé reads paN+Da ka, and S reads pAN+Da ra.

n.259Tib. snying por ’gro ba. This nāga king is not attested in the Sanskrit version.

n.260S includes another nāga king who could not be identified: klu’i rgyal po ri brtsegs.

n.261Tib. klu’i rgyal po srin bu, the Sanskrit of which is attested in the Mahāvyutpatti. This line is not attested in the Sanskrit version of this text.

n.262Tib. e la’i ’dab. This line is not attested in the Sanskrit.

n.263Tib. dkar po. This line is not attested in the Sanskrit, and there are many possible Sanskrit equivalents for dkar po.

n.264Tib. po la ba. The name of this nāga king is derived from the transliterated Sanskrit given in the Tibetan text. The Sanskrit reads kambala here.

n.265Gaṅgā, Sindhu, Vakṣu, and Sītā are the names of rivers. They are commonly identifed as the Ganges, Indus, Oxus, and Tarim rivers, respectively. See also 1.234 below, where these same four rivers are identified as “river queens.”

n.266This is the second instance of this name in the Tibetan translation.

n.267This follows the Sanskrit spelling. The Tibetan reads gtsug phud lha, which should perhaps be emended to gtsug phud lnga.

n.268Tib. stobs can. This line is not attested in the Sanskrit.

n.269Tib. tsam pa ka. This line is not attested in the Sanskrit.

n.270Tib. yid bzang. This line is not attested in the Sanskrit. S includes a line that reads klu’i rgyal po mi bzang, which could be understood as “the nāga king Sumānuṣa.”

n.271The Tibetan reads me, which has been emended to mi following the Sanskrit and S.

n.272Tib. gzings. This line is not attested in the Sanskrit.

n.273This line is not attested in the Sanskrit.

n.274Degé reads a lu ka but has been emended based on the Sanskrit and S (u lu ka).

n.275The Sanskrit reads luka.

n.276Tib. e la’i lo ma. The Sanskrit reads elavarṇa.

n.277The names of the previous two nāga kings follow the Sanskrit, as the Tibetan names are difficult to interpret. They are ’gro sgra and ’chi sgra, respectively.

n.278This line is not attested in the Sanskrit.

n.279Reading mgyogs rab instead of mgyogs rab dga’, which appears to be a scribal error in D.

n.280The names Parikāla and Parikīṭa follow the Sanskrit. The Tibetan reads mchog can and mchog ldan grog ma, respectively.

n.281This follows the Sanskrit. The Tibetan reads seng ge ’dzin.

n.282As above, this translation follows the Sanskrit in reading kṛtyākarmaṇa as a single term, rather than as two terms as they appear in the Tibetan, gshed byed dang sgyur ba’i las.

n.283The Sanskrit has the following mantra: araḍe karaḍe maraḍe madavardhane avaśabare ture ture care care śabare parṇaśabare hucu hucu kuci kuci muci muci huci huci huci huci huci huci huci muci muci muci muci svāhā.

n.284Sanskrit has the following mantra: iḍi miḍi khire vikhire hili hili hili mili mili tumūle ambare ambarāvati dumbe dumbe hili hili hili kurvi kurvi kurvi muci muci muci svāhā.

n.285Sanskrit has the following mantra: morike morike morike kevaṇṭu maṇḍitike hare hare hare hare hare hare ghare ghare khare khare phare phare phare phali nidanti nidantire śakaṭi makaṭi naṇḍa naḍini śiri śiri śiri śiri śiri śiri svāhā.

n.286Sanskrit has the following mantra: hiḍi miḍi kuḍi miḍi tuḍi kuḍi aḍi muḍi tuḍi tuḍi āḍe dante dantile śakari makāri thagari cagari kañcane kañcanāvati śabare bare bare bare bare bare bare dante siddhi svāhā.

n.287Sanskrit has the following mantra: tantule tatale tatale calate tale tale tale tale vire vijaye vijjadhare araje araje virajāmati mati mālini maṇḍe śirimaṇḍe jvala jvala jvala jvala jvala jvala bhadravati siddhi svāhā.

n.288This is followed by paṇḍare in the Sanskrit.

n.289Siddhi follows the Sanskrit, C, H, K, Y, and N. D reads siddhe.

n.290Sanskrit has svāhā siddhi siddhi siddhi siddhi svāhā instead of siddhi svāhā, as in the Degé.

n.291Sanskrit has the following mantra: hili mili kili mili ilile katale ketumūle aḍamali ḍaphe ḍaphe busareke busaṭṭa narakande katili bocalike kambu dāriṇikāya śabaraṅge tatuṭṭa bharaśebho vardhani prakṛti daṃṣṭra mili tale īti hāse amale makule baṭṭa baṭṭa.

n.292Sanskrit has the following mantra: iṭiṭāya godohikāya bhṛṅgārikāya aruci maruci naṭṭa naṭṭa vajre vajranaṭṭe udayanapriye ale tāle kuntāle kule tāle nārāyaṇi pārāyaṇi paśyani sparśani.

n.293This line presents a challenge in having what appears to be two different first-person referents. Śākyamuni, as the Thus-Gone One, is referred to using the first-person instrumental pronoun bdag gis (Skt. mayā), followed by the first-person dative referent bdag la, to refer to the person reciting this passage. There is no Sanskrit equivalent for bdag la, as it reads “all sentient beings” (sarvasattvānām) instead. It seems unlikely, but not impossible, that Śākyamuni is the referent in both cases.

n.294Sanskrit adds haraṇi.

n.295Sanskrit, Y, K, L, and C have bodhi bodhi instead of bedhi bedhi.

n.296Sanskrit has instead bodhiparipācaṇīye.

n.297Sanskrit has the following mantra: hili hili hili hili mili mili mili mili mālini caṅkali kiri kiri kiri kiri kiri kiri kiraye brahmāye ratnakaraṇḍake viḍāphu viḍāphuse dhara dhara hara hara hule hule phuru phuru phuru phuru phuru svāhā.

n.298This follows the Sanskrit in reading brahmādaṇda. The Tibet has the ambiguous phrase tshangs pa’i me dbal. Elsewhere in this text me dbal is used to translate vaisarpa (“boils”).

n.299Tib. dogs pa’i dug, which translates the Sanskrit śaṅkāviṣa. The Carakasaṃhitā, a foundational treatise of Āyurvedic medicine, describes śaṅkāviṣa as the occurrence of psychosomatic symptoms of poisoning in someone who believes they have been poisoned but have not (6.23.221–23; p. 481). The Sanskrit version of The Great Peahen reads śakraviṣād, “poison from power” or “poison from Śakra” (i.e., Indra). This likely a scribal corruption of śaṅkāviṣa.

n.300Sanskrit has the following mantra: jalā jantule cāpeṭi jantule mathani ghaṭani grasani hari hari śiri dyoti śire tataru tataru ṇabati siṃhā hā hā hā hā hā siṃhe dhiti dhiti kuru kuru śabare vajre jyoti tuṭa tuṭasi baṭa baṭasi sili sili kapile kapilamūle hā hī hūm.

n.301Tib. sum bcu pa’i lha. The Sanskrit reads tridaśehi devehi, which could be more literally interpreted as “thirty gods.”

n.302Sanskrit has the following mantra: ḍaṭaṅgini surapativarti vajra vajra vajra vajra vajra vajra vajrapataye svāhā.

n.303Sanskrit has the following mantra: jvala jvalana tapa tapana dhama dhamana śara śaraṇa nakuṭi nakuṭi muṭi muṭi sara sara mara mara hara hara bhara bhara titi tiri ṭa ṭa ṭa ṭa ṭa dā dā dā dā dā vā vā vā vā vā hale hale hale hale hale siddhi siddhi siddhi siddhi siddhi svasti svasti svasti svasti svasti mama sarva­sattvānāṃ ca svāhā.

n.304Skt. kālapāśā. This term generally refers to the noose wielded by Yama, but because it is given here in the feminine it is possible it refers to the noose in the form of a deity.

n.305The Tibetan renders these names in the masculine as “kings.” Given their feminine gender of nadīrajñī (“river queens”) in Sanskrit, and given the Indian custom of associating rivers with feminine divinities, we have followed the Sanskrit. As in previous sections, we have drawn from the attested Sanskrit river names when the meaning of the Tibetan term was ambiguous.

n.306The Tibetan reads zhi ba ’dzin, which does not appear to be equivalent to the extant Sanskrit śatabāhu .

n.307This follows the Sanskrit. The Tibetan reads chu mchog can.

n.308The Degé reads ’phags pa can, but this seems to be in error. S reads pags pa can, which aligns with the Sanskrit carma, meaning “skin.”

n.309This follows the Sanskrit. The Tibetan reads me tog pad+ma.

n.310Tib. klung chen. This river is not attested in the Sanskrit.

n.311Tib. ba lang sbyin mchog can. This river is not attested in the Sanskrit.

n.312The Tibetan reads gsang srung, which would suggest the Tibetan translators read Rahasyā. However, Rathasyā, as a variant spelling of Rathasthā, is an attested name of a river in Indic literature. S reads shing rta’i bdag, which aligns closely with the Sanskrit.

n.313Here the Tibetan includes another river queen with the name ba lang can, which is very similar to Gomatī (ba lang ldan), above.

n.314As in previous sections, we have drawn from the attested Sanskrit names where the meaning of the Tibetan term is ambiguous.

n.315This follows the Sanskrit. The Tibetan reads khyab gnas.

n.316This follows the Sanskrit. Degé reads tshangs pa’i mchu; H, N, and S read tshangs pa’i chu.

n.317The Sanskrit reads arjana, which has been emended based on the Tibetan mig bsku.

n.318Tib. muny+dza. This mountain is not attested in the Sanskrit.

n.319The Sanskrit reads upātiṣṭi, which has been emended based on the Tibetan nye dkar.

n.320This follows the Sanskrit. The Tibetan reads stobs ’dzin.

n.321The Tibetan includes a mountain king named dus ’bri byed, whose name is difficult to interpret but could be translated tentatively as “causes time to be written” or “causes time to diminish.” This name does not correspond to any name in the Sanskrit or to any attested name of a mountain.

n.322The Sanskrit reads śaśananāda, which has been emended based on the Tibetan bstan pa ’dzin.

n.323This follows the Sanskrit. The Tibetan reads lha mtshams.

n.324The term “star” (tārā; skar ma) refers to the seven planets, Rāhu (the eclipse), and Ketu (meteors/comets). The sun and moon are excluded. The Sanskrit edition reads, “The five stars, with Rāhu and Ketu, are in that way seven” (tārāgrahās tathā pañca rāhuketuś ca saptamaḥ).

n.325As in previous sections, we have drawn from the attested Sanskrit names where the meaning of the Tibetan term is ambiguous.

n.326The Sanskrit reads balamīka, which has been emended based on the Tibetan grog mkhar ba.

n.327This follows the Sanskrit. The Tibetan reads ’gro ’dzin.

n.328This follows the Sanskrit. The Tibetan reads ’gro dga’.

n.329This follows the Sanskrit. The Tibetan reads ’jigs can.

n.330The Sanskrit reads yamadagni, which aligns with the Tibetan translation gshin rje’i me. However, the name of this ṛṣi is more typically spelled jamadagni , which has been followed here.

n.331This follows the Sanskrit. The Tibetan reads ngag mnyam.

n.332This follows the Tibetan. The Sanskrit includes two names here that do not seem to have equivalents in the Tibetan translation: śarabha and mardana .

n.333This follows the Sanskrit. The Tibetan reads cha byad ’dzin.

n.334This follows the Sanskrit. The Tibetan reads gyur mid kyi bu (S: khyur mid kyi bu).

n.335This follows the Sanskrit. The Tibetan reads gar ldan gyi bu.

n.336This follows the Sanskrit. The Tibetan reads sgra’i bu.

n.337In the Sanskrit this name is preceded by three names for which there is no equivalent in the Tibetan translation: sunetra , suranemin, and asita.

n.338The Sanskrit has the following mantra: hiri hiri hiri khiri khiri miri miri muri muri hiri hiri mili mili ḍaphu ḍaphu ḍaphu mathani ghāṭani pacani pācani dahani dāhani tapani tāpani hanani daha daha dahani tālani pātani mojani sta stani jastani svayambhūve svāhā.

n.339The Tibetan term dge ba translates two different Sanskrit terms in this line: śubha (“virtue”) and svāsti (“fortune”).

n.340The following list relies heavily on the Sanskrit text, as many of the Tibetan names have no clear equivalent in Sanskrit. There are many sets of prajāpati s in Sanskrit literature, but this list of twelve appears to be unique.

n.341The Sanskrit reads atṛ, which has been emended to atri , the standard spelling of the name of this well-known mahāprajāpti.

n.342The Sanskrit has the following mantra: hiri hiri khiri khiri mili mili sili sili hili hili mili mili sili sili ḍaphu ḍaphu ḍaḍaphu grasani mathani dahani ghātani pacani pācani hanani dāhani dala dala dālani pāṭani mohani sta stanī jambhanīye svāhā.

n.343Apart from a few minor emendations, the following list of poisons is rendered as it appears in transliterated Sanskrit given in the Degé version of the text. There are numerous orthographic variations across the Tibetan versions, as well as differences in spelling and number between the Tibetan and Sanskrit versions.

n.344This follows the Sanskrit. The Tibetan reads ser skya, suggesting the Tibetan translators read kapila.

n.345This follows the Sanskrit. The referent of the Tibetan term kyab gnas could not be identified.

n.346The follows the Tibetan pad+ma dkar po. The Sanskrit has udumbala here.

n.347This follows the Sanskrit. The referent of the Tibetan term ser po could not be identified.

n.348This follows the Sanskrit. The referent of the Tibetan term rgyas byed could not be identified

n.349This reading follows the Sanskrit, F, H, K, Y, N, and S in reading avadhūtaº (smad pa rnams). Degé reads gnod pa rnams (“harms”).

n.350This follows the Sanskrit and S in reading śastra (mtshon cha, “weapons”), where the Degé reads gsangs sngags (“mantra”).

n.351The Sanskrit reads balaki. N has dhāraki, and K and Y read dharaki.

n.352Sanskrit has kuru tulu rather than hulu hulu as in the Degé. K, Y, and N have hulu tulu.