Introduction

i.1The Noble Dhāraṇī “The Invincible Sitātapatrā Born from the Uṣṇīṣa of the Tathāgata” (Toh 592) is one of four texts preserved in the Degé Kangyur (Toh 590–93) dedicated to the female deity Sitātapatrā (gdugs dkar po can), the White Umbrella Goddess. Though these four texts differ somewhat in length and arrangement, they all share the same core material and thus represent four unique variations of a single work. At the heart of each of these texts is a series of spell formulas that can be recited to avert a wide array of threats to health, well-being, and prosperity. The spell of Sitātapatrā has enjoyed sustained popularity as a source of security and protection in numerous Buddhist communities, as evidenced by its long and complex textual history and the numerous languages into which it has been translated. The four texts translated into Tibetan and preserved in Kangyur reflect distinct stages of the spell’s evolution, stages that mirror its development in the broader Buddhist community. Toh 592 is the shortest of the four canonical translations and may be the earliest of the four versions translated into Tibetan, thus representing a relatively early stage in the spell’s evolution in the Indic Buddhist tradition.

i.2Toh 592 is unique among the four canonical translations of the Sitātapatrā texts for omitting the scriptural introduction (nidāna; gleng gzhi) that sets the stage for Śākyamuni’s revelation of the deity and her spell. In the three other versions, the text begins in the Heaven of the Thirty-Three, where Śākyamuni is resting in samādhi among an assembly of monks, bodhisattvas, and the gods of the realm. While he is deep in samādhi the spell issues from his uṣṇīṣa, resounding in full throughout the assembly. It begins with a long series of homages to the Three Jewels, an array of buddhas and other realized beings, and a number of gods and other figures from the brahmanical pantheon, including Brahmā, Indra, Śiva, and Viṣṇu. This opening homage is followed by verses invoking Sitātapatrā in the form of various female deities, including Tārā, Bhṛkuṭī, and Pāṇḍaravāsinī, thereby equating her with many of the most renowned female deities of the Buddhist tradition. Most of the teaching is dedicated to a series of spells and other recitation formulas that enjoin Sitātapatrā to intervene on the practitioner’s behalf to avert an exhaustive list of diseases, afflictions, rival spells, and the adverse influences of supernatural beings. The text concludes with a description of the effectiveness of the spell and the benefits of relying on Sitātapatrā.

Sitātapatrā and Her Spell

i.3Sitātapatrā is at once the name of a spell and the deity it invokes. In the title of Toh 590 and throughout all four texts, Sitātapatrā is called a vidyā , a term that refers to both a class of deities and a type of magical formula, thus indicating their inseparability. To recite Sitātapatrā’s spell‍—or to wear it, inscribe it on a talisman, insert it into a caitya, and so forth‍—is to summon the powerful deity to intercede on one’s behalf. The primary name of the spell in Sanskrit is sarva­tathāgatoṣṇīṣa­sitātapatrā, which is somewhat ambiguous given that the precise relationship between the compound sarva­tathāgatoṣṇīṣa (“uṣṇīṣa[s] of all tathāgatas”) and sitātapatrā can be read in a number of plausible ways. The Tibetan translators settled on a specific interpretation by inserting the phrase nas byung ba (“born from”) in all versions of the title so that it reads, in Tibetan translation, Sitātapatrā born from the uṣṇīṣa of all tathāgatas . As this aligns well with the setting of the sūtra, in which the spell emerges from Śākyamuni’s uṣṇīṣa, we have followed this interpretation here.

i.4As a magical formula, Sitātapatrā born from the uṣṇīṣa of all tathagatas is alternatively referred to as a vidyā (“spell”), a mahāvidyārajñī (“great queen of spells”), a dhāraṇī , and a mantra. These terms are used interchangeably to refer to the magical formulas that are used to avert the threats of disease, misfortune, aggression, and the influence of supernatural beings. Because the spell is held to be specifically effective for averting these threats before they strike, the spell is designated a pratyaṅgirā, an “averting” or “counter” spell. And, because it is regarded as highly potent for this purpose, it is further referred to as aparājitā (“invincible”).

i.5The dangers Sitātapatrā can capably avert are enumerated in great detail and include a litany of physical illness and mental disorders, a vast demonology of supernatural forces that cause illness and distress, threats from kings, poisons, and animals, and even a detailed list of rival magical traditions whose spells pose a potential threat. Given this exhaustive treatment of the benefits of the spell, it is noteworthy that the path to liberation and the attainment of buddhahood are never mentioned. While it can be implicitly understood that averting disease, calamity, and supernatural dangers are requisites for the pursuit of awakening, spiritual goals are clearly subordinated in these texts to the goal of alleviating the worldly anxieties shared by all beings, Buddhists and non-Buddhists alike.

i.6The texts on Sitātapatrā preserved in the Kangyur do not provide a detailed iconography of the goddess, saying only that she has a thousand heads, a thousand arms, a thousand legs, and a trillion eyes. This form of Sitātapatrā is still popular in the contemporary Buddhist tradition, but she is also depicted in a number of other forms in the various practice manuals derived from the canonical texts. This includes forms of the goddess with one face and two arms (Toh 3084), three faces and six arms (Toh 3114), and five faces and eight arms (Toh 2689).

i.7The circulation of texts on Sitātapatrā can be traced back to at least the eighth century, which is the proposed date of the earliest textual witnesses available. Given that the earliest versions of the spell were discovered in Central Asia, it is clear the spell was popular well before this time. Sitātapatrā continues to be relevant in the contemporary Vajrayāna traditions of Buddhism, especially in Nepal and Tibet, as demonstrated by the numerous copies of her spell that circulate. In Tibet, the Sitātapatrā spell was widely popular from an early period, as indicated by the large number of Sitātapatrā texts discovered at Dunhuang. A version of the Sitātapatrā spell is also said to have been specifically translated for Tri Songdetsen (khri srong lde’u btsan, r. 756–800), as we find it included among the “ten royal sūtras” (rgyal po’i mdo bcu) translated for the king at Padmasambhava’s recommendation. Numerous practice manuals and ritual texts for Sitātapatrā have been composed in Tibet into recent times, many of which draw explicitly from the canonical sources.

The Canonical Texts

i.8The four Sitātapatrā texts preserved in the Degé Kangyur are classified as kriyātantras, and they are further categorized among texts associated with the tathāgata family and listed alongside texts associated with other uṣṇīṣa deities such as Uṣṇīṣavijayā. As is often the case with spells and dhāraṇīs, the Sitātapatrā spell is also included in the Dhāraṇī Collection (gzungs ’dus) of the Degé Kangyur as Toh 985 and 986, which correspond to Toh 590 and 592. The four canonical texts (Toh 590–593) represent four distinct versions of the same spell that are largely equivalent in terms of content, translation style, and terminology. Two of the four lack a colophon describing the context of their translation, but it is nonetheless apparent that the later versions of the text are in fact revisions of earlier Tibetan translations based on newly-available Sanskrit sources rather than distinct translations. Though many ambiguities remain, the four works offer us an important view into the long textual history of both the Indic source material and its Tibetan translations.

i.9Toh 590, The Noble Invincible Great Queen of Spells for Averting Called “Sitātapatrā Born from the Uṣṇīṣa of All Tathāgatas” (Ārya­sarva­tathāgatoṣṇīṣa­sitātapatrā­nāmāparājita­pratyaṅgirā­mahāvidyārājñī; ’phags pa de bzhin gshegs pa thams cad kyi gtsug tor nas byung ba gdugs dkar po can zhes bya ba gzhan gyis mi thub ma phyir zlog pa’i rig sngags kyi rgyal mo chen mo), the longest of the four and most closely aligned with the more recent Sanskrit witnesses, lacks a translator’s colophon, so it is impossible to determine its date, but its length and its similarity to the later Sanskrit manuscripts suggests that it is the most recent of the versions in the Kangyur. A unique, alternative translation of the text corresponding to Toh 590 is preserved in the Phukdrak (phug brag) Kangyur. This translation, which was made by the eleventh-century Indian paṇḍita Vibhūticandra and the Tibetan translator Sherap Rinchen (shes rab rin chen), is a revision of Toh 590 based on additional Sanskrit manuscripts not available to the anonymous translator of Toh 590. Toh 590 was also revised or retranslated in the fifteenth century by Sönam Nampar Gyalwa (bsod nams rnam par rgyal ba; 1401–75) of Jampa Ling monastery (byams pa gling) in Central Tibet. His translation, which is available only in his collected writings, was based on his own study of Indic manuscripts and consultation with the Burmese Buddhist paṇḍita Alaṅkāraśrī of Haṃsāvati (Pegu).

i.10Toh 591, titled The Noble Dhāraṇī “The Supreme Accomplishment of Invincible Averting, Sitātapatrā Born from the Uṣṇīṣa of the Tathāgata” (Ārya­tathāgatoṣṇīṣa­sitāta­patrāparājita­mahāpratyaṅgira­paramasiddhanāma­dhāraṇī; ’phags pa de bzhin gshegs pa’i gtsug tor nas byung ba’i gdugs dkar po can gzhan gyis mi thub pa phyir zlog pa chen mo mchog tu grub pa zhes bya ba’i gzungs), is shorter than Toh 590 and in this regard is perhaps closer in content to Toh 592 and 593 in lacking many of the lines in the opening homage found in Toh 590, but it nonetheless represents a distinct arrangement of the material in dividing the verse section listing the names and epithets of the goddess into two sections interspersed with one of the spell formulas. It is also unique for designating two of the spell formulas as “essence mantra” (snying po) and “subsidiary essence mantra” (nye ba’i snying po), designations that are not found in any of the Sanskrit sources consulted. Toh 591 identifies itself not as a translation but as a revision of a prior Tibetan translation. The colophon does not use the verb “translated” (bsgyur), but instead tells us that this version, prepared by the Kashmiri master Parahitabhadra (ca. eleventh century) and the Tibetan translator Zu Gawé Dorjé (gzu dga’ ba’i rdo rje), is based on a comparison of a prior translation with an “old” manuscript discovered at the Amṛtabhavana monastery in Kashmir. Though the prior translation that served as the basis for the revisions of Toh 591 cannot be definitively identified, it seems probable that the version was either Toh 592, Toh 593, or a version similar to those translations.

i.11Toh 592 and 593, both of which are titled The Noble Dhāraṇī “The Invincible Sitātapatrā Born from the Uṣṇīṣa of the Tathāgata” (Ārya­tathāgatoṣṇīṣa­sitātapatrānāmaparājitānāma­dhāraṇī; ’phags pa de bzhin gshegs pa’i gtsug tor nas byung ba’i gdugs dkar po can gzhan gyis mi thub pa zhes bya ba’i gzungs), are nearly identical versions of the Sitātapatrā spell. Toh 592 lacks the scriptural introduction and conclusion found in Toh 593 but otherwise varies only slightly and in a manner more consistent with scribal errors and editorial interventions than differences in the source material. Toh 592 lacks a translator’s colophon, making it difficult to determine its origin, but a text with nearly the same title is recorded in the Denkarma (ldan dkar ma), the imperial-period register of Tibetan translations. Toh 593, which does include the introductory and concluding passages absent in Toh 592, has a colophon reporting it to be a translation by the eleventh-century Kashmiri master Mahājana made without the assistance of a Tibetan translator. It is therefore possible that Mahājana’s contribution to the collection was to add the introductory and concluding material known from other Sitātapatrā sources. Mahājana’s colophon identifies the text as a “version of the Uṣṇīṣa ” that is “the shorter of those of the heavenly realm.” This ambiguous statement is made somewhat clearer in the catalog of the Urga Kangyur, which says that Toh 593 (Urga no. 594) “is renowned as the shorter Uṣṇīṣa of the heavenly realm” (lha yul ma chung bar grags pa). Sönam Nampar Gyalwa (see i.­9 above) also refers to this translation as the “condensed version” (bsdus pa) of the Sitātapatrā spell.

i.12A comparison of the four canonical translations of Sitātapatrā’s spell suggests that they represent three distinct branch recensions of the same source material and thus reflect the evolution of the text in the Indic tradition. Toh 590 and 591 constitute two of those branches, while Toh 592 and 593 together represent the third. This was the view of Sönam Nampar Gyalwa, who offered this statement about the relationship between the texts in the colophon to his own revision of Toh 590:

There are three versions of this dhāraṇī rite. The most extensive is this text, The Great Queen of Vidyās (Toh 590), for which the previous translator is unidentified. The middle-length version is the one known as The Supreme Accomplishment (Toh 591), which was translated by Zu Gawé Dorjé. The concise version is [called] “the one known as the lesser of the heavenly realm” (Toh 593) and was translated by the Kashmiri paṇḍita Mahājana. There is another, shorter version of “the one known as the lesser of the heavenly realm” (Toh 592) that is distinct only for lacking the scriptural introduction. It need not be counted [separately].

i.14This brief survey of the four canonical translations allows for a tentative argument to be made about the translation and propagation of this series of Sitātapatrā spells in Tibet. The spell was likely first translated during Tibet’s imperial period, as indicated by the two imperial-period catalogs, the Denkarma and Phangthangma (phang thang ma). Whereas the title of the text in the Denkarma, ’phags pa de bzhin gshegs pa gtsug tor nas byung ba’i gdugs dkar mo can gzhan gyis mi thub pa, aligns closely with that of Toh 592/3, the title in the Phangthangma, ’phags pa gtsug tor gdugs dkar po, is generic and thus could refer to any of the four canonical texts, or to a different, unknown version. It is possible that the earliest version of the four canonical texts is Toh 592, which lacks a colophon indicating its provenance. If this is the text recorded in the Denkarma it would have been translated no later than 843, the year the Tibetan empire collapsed and record of its translation efforts ceased. Toh 593, which does include a colophon dating it to the eleventh century, represents the same branch recension but, as noted above, differs in its inclusion of the introductory and concluding statements‍—perhaps Mahājana’s specific contribution to the corpus. Toh 591, which is described in its colophon as a revision rather than a new translation, was also prepared in the eleventh century. It differs only slightly from Toh 592/3, primarily in its unique arrangement of the material. Thus it appears that Toh 591 and 593 comprise a second period of translation of the Sitātapatrā spell in the eleventh century, one in which the earlier translation represented by Toh 592 served as a primary point of reference. It is especially noteworthy that this second wave primarily involved Indian masters and manuscript witnesses from Kashmir. Thus, Toh 590 is likely the last of the translations to be produced, and then was revised two additional times as described above.

Other Sources

i.15As noted above, the widespread popularity of Sitātapatrā is attested by the broad circulation of the Sitātapatrā spell. Numerous versions are preserved in Sanskrit, Khotanese, Chinese, Old Uyghur, and Tibetan, thus demarcating its circulation throughout South Asia, the Himalayan region, Central Asia, and China. What is perhaps the oldest documented Sanskrit witness of the spell, tentatively dated to the eight century, was discovered at Dunhuang and written in a Gupta script unique to Central Asia. This version, published in 1963 by H. W. Bailey, was consulted for this translation. The popularity of Sitātapatrā in the Newar Buddhist tradition is evident in the large number of extant Sanskrit manuscript witnesses of the spell scribed in Nepal. Many of these versions are found in the numerous dhāraṇī collections (dhāraṇīsaṅgraha) popular in the Newar tradition. Most of the available manuscripts are relatively recent, dating no earlier than the eighteenth century. A representative collection of Nepalese manuscripts was consulted for this translation, the most noteworthy version of the spell being found in Cambridge Ms. Add 1326, a dhāraṇīsaṅgraha compiled in 1719. This version, like most Nepalese versions consulted, most closely aligns with Toh 590.

i.16Also noteworthy are the versions of the spell composed in Old Uyghur, which were translated from an unknown source language in likely the thirteenth or fourteenth century. The manuscripts were discovered in Turfan in the early twentieth century and are now dispersed among various European and Russian manuscript archives.

i.17There are two Chinese translations of works that are similar in title and content to Toh 590, but a close comparison of the Tibetan and Chinese translations is needed to determine precisely how the two Chinese translations and four Tibetan translations align. Taishō 976, Fo ding dabai sangai tuoluoni jing (佛頂大白傘蓋陀羅尼經), was translated by the Tangut monk Shaluoba (1279–1314), and Taishō 977, Fo shuo dabai sangai zong chi tuoluoni jing (佛說大白傘蓋總持陀羅尼經), was translated by Zhen Zhi sometime during the Yuan period (1271–1368). Based on these dates it would appear that both Chinese translations significantly postdate the Tibetan translations preserved in the Kangyur.

i.18Finally, there were a number of Tibetan versions of the Sitātapatrā spell discovered among the Dunhuang manuscripts. These versions are revealing in that they are shorter and otherwise distinct from the canonical versions, indicating one or more additional branch recensions. Some of the Dunhuang manuscripts do align with Toh 592, the canonical version proposed as the earliest, but none appear to correlate directly with Toh 590 and 591, which are believed to have been translated in or after the eleventh century, long after the Dunhuang caves had been sealed.

The Translation

i.19The present translation is based on the Tibetan version in the Degé Kangyur, in consultation with the Stok Palace and Phukdrak versions as well as the variant readings recorded in the Comparative Edition (dpe bsdur ma) Kangyur. Extensive use was made of Sanskrit witnesses, including the Khotanese version and four representatives from the numerous Nepalese manuscript witnesses. Among those, Cambridge Ms. Add. 1326 and its edition prepared by Gergely Hidas proved especially useful for resolving ambiguities in the Tibetan translation and correcting minor but consequential orthographic errors in the Tibetan transliterations of Sanskrit spell formulas. Apart from those necessary corrections, the spell formulas follow the transliterations presented in the Degé version. Even with the wealth of resources available, a number of enigmatic passages remain imperfectly resolved, particularly in the verse section recounting the names and epithets of the deity. Tentative translations of these difficult passages have been offered, but they are not intended to represent a definitive interpretation.