Kathmandu, Calcutta, Paris: the life of a Tantric manuscript

Scholars have long sought to classify people and things based on their geographic origin. In the Muqaddimah, Ibn Khaldun insisted that climate shapes both the appearance and temperament of different human races; Vātsyāyana imagined an entire geography of desire in his Kāmasūtra. Space, as well as time, seem to be natural measures with which we partition human culture into different orders. This categorical impulse has naturally extended to the study of manuscripts — scholars painstakingly sort minute differences in script, material, and format into various categories and precisely-qualified subcategories to try and situate a manuscript in a specific place, to narrow it down to a specific era.

But people move; they criss-cross geographical and political boundaries; they dip in and out of the cultural spheres dreamed up by anthropologists. And as they travel, they bring their manuscripts, traditions, and knowledge with them. Sometimes, this results in hybrid artefacts that are so singular as to be nigh uncategorizable. But, as scholars, we have learned to see things through the lens of our expertise; our categories are our eyes. And the curious history of some Tantric fragments at the BnF is a perfect case study in what happens when they fail us.


To Nepal and back


Haraprasāda Śāstri in 1928 (Proceedings and Transaction of the Fifth Indian Oriental Conference, 1930, 62 bis).

Haraprasāda Śāstri circa 1928 (Proceedings and Transaction of the Fifth Indian Oriental Conference, 1930, 62 bis).

Haraprasāda Śāstrī (1853-1931) was one of the most important Bengali scholars of the turn of the 20th century. Equally eloquent in English and Bengali, he published prolifically on many topics, but perhaps he would have liked to be remembered as a Sanskritist.1 When the great Rāja Rājendralāla Mitra died, H. P. Śāstrī picked up where he left off, collecting Sanskrit manuscripts for the Bengal presidency from all over India. But whereas Rājendralāla Mitra’s search was preoccupied with Vedic manuscripts, H. P. Śāstrī had become increasingly interested in Tantra. In the preface to the first Notices of Sanskrit Mss published by him in 1892, he puzzles over the relationship between Vedic and Tantric traditions, and laments the lack of a “systematic attempt” to collect Tantric works.

In the winter of 1884-85, Cecil Bendall, a cranky Cambridge professor at Caius College, had made a trip to Nepal and caught a brief glimpse of the library of the Mahārāja. He had been greatly impressed by the collection and was itching to go back. He had learned of the Mahārāja’s collection from Daniel Wright, who had doubted its existence, but who had nevertheless sent back a number of rare manuscripts from Kathmandu. Among them was a fragment of the Pārameśvaratantra in late Gupta script, with a colophon dated to the 9th century, the oldest dated Sanskrit manuscript in Cambridge.2 Bendall, however, was only able to spend four hours at the library, and had to leave before being able to obtain any manuscripts. From Nepal, he continued on to Calcutta, where he — although finding no manuscripts worth collecting — made the acquaintance of H. P. Śāstrī, who had become known as a “collaborateur” in Rājendralāla Mitra’s Sanskrit Buddhist Literature of Nepal, translating the descriptive catalogue entries prepared by his paṇḍits from Sanskrit into English.

In 1897, H. P. Śāstrī decided to visit the library in Kathmandu for himself, in what would prove to be the first of many such journeys. He went for a second time in 1898, in the company of Bendall and Binodavihārī Bhaṭṭācārya, and, to his delight, the library was full of Tantric manuscripts; he later relates that the librarian, Subbā Viṣṇu Prasāda Rājabhāṇḍārī, “was a Tantrika himself and passed his time in reading Tantras.” During this visit, H. P. Śāstrī found a number of interesting texts, including one startling discovery — a palm-leaf manuscript of the Kulālikāmnāya, a Tantric work on the worship of the goddess Kubjikā. Comparing it palæographically with the Hōryūji manuscript from Japan, he declared it to be written in “early Gupta” script and thus from the 7th century at the latest, even older than the Pārameśvaratantra manuscript in Cambridge. He acquired the Kulālikāmnāya manuscript for the Asiatic Society of Bengal and brought it back with him to Calcutta, and, in his descriptive catalogue of Tantric manuscripts at the Asiatic Society — published posthumously —, it takes pride of place as its first catalogue entry.

Other scholars picked up on this discovery — John Nicol Farquhar, a Scottish missionary living in Calcutta, wrote in 1920 that the Kulālikāmnāya manuscript proved that “Śākta theology and ritual were already in existence about A.D. 600,” a claim that was cited by Jan Gonda in 1960.3 No one seems to have noticed that H. P. Śāstrī changed his mind; toward the end of his life, he spoke of the manuscript as being “written in the 8th or 9th century”, and, in the posthumous catalogue entry, the script is described as “Later Gupta”.


A palæographical mystery, solved


In 1971, Karel Rijk van Kooij published a detailed study of the Kulālikāmnāya manuscript, comparing it to other manuscripts — including the Cambridge Pārameśvaratantra and the Hōryūji manuscript — from various times and places. In contrast to H. P. Śāstrī, van Kooij dates the manuscript to the 12th century and concludes that it is, in fact, written in an early form of Śāradā script, which is typically used in Kashmir.


van Kooij 1971, 882. Courtesy of the Martin-Luther-Universität Halle-Wittenberg.

van Kooij 1971, 882. Microfilm image showing folios 12v and 13r of the Kulālikāmnāya manuscript. Courtesy of the Martin-Luther-Universität Halle-Wittenberg.

This is not something that H. P. Śāstrī would have expected at all. The appearance of Śāradā script in Nepal is unusual, and, written on long palm leaves, doubly so. Śāradā manuscripts from Kashmir are typically written “portrait”-style, either on birch bark or on paper. As Michael Witzel points out, the existence of a palm-leaf Śāradā manuscript in Nepal seems to indicate that “a Kashmiri Tantric came to Nepal and brought the MS along or that he even copied it while staying in Nepal.4 This scenario was perhaps so unlikely to H. P. Śāstrī that it did not even cross his mind. His expertise was, predominantly, in Bengali and Nepalese manuscripts; when he saw the familiar palm leaf format, as well as writing that recalled that of the dated “Gupta”-style scripts, he naturally concluded that this manuscript should also fall into the same category.


Birch barck manuscript from Kashmir. Wellcome collection. The original image is displayed upside-down.

A birch bark manuscript from Kashmir. Wellcome collection. The original image is displayed upside-down.


A French doctor comes to town


Palmyr Cordier in 1894. Reproduced from Filliozat 2014, originally from Roșu 1989.

Palmyr Cordier in 1894. Reproduced from Filliozat 2014, originally from Roșu 1990.

In 1897, the same year that H. P. Śāstrī first set off for Nepal, Palmyr Cordier (1871-1914), a French doctor and Orientalist, had been dispatched to serve in the colony of Chandernagor, just north of Calcutta. He, too, was very interested in Sanskrit manuscripts, especially of medical texts, having written his doctoral thesis on Hindu medicine. When H. P. Śāstrī returned to Calcutta after his second trip to Nepal, he met Cordier and showed him a number of Nepalese manuscripts of Ayurvedic texts, which Cordier photographed. From those photographs, which Cordier described as being clearer and easier to read than the manuscripts themselves, he had a scribe — often an Ayurvedic practitioner — create a fresh paper copy.5 These copies are full of annotations in Cordier’s hand; when he finally published a preliminary report on the manuscripts, he offered a number of corrections to H. P. Śāstrī’s catalogue descriptions.


BnF Sanscrit 1155 Kāśyapasaṃhitā, cover. Courtesy of the BnF. Manuscript copied from photographs of a Nepalese manuscript of H. P. Śāstrī, now lost. See Wujastyk 1998, 207. The colophon indicates that this copy has been made by Kavirāja Umeśacandra Kaviratna, an Ayurvedic practitioner that Cordier met in Chandernagor. H. P. Śāstrī later refers Julius Jolly to Cordier for help obtaining copies of this manuscript, although Cordier had already left Calcutta by then.

BnF Sanscrit 1155 Kāśyapasaṃhitā, cover. Courtesy of the BnF. Manuscript copied from photographs of a Nepalese manuscript of H. P. Śāstrī, now lost.6 The colophon indicates that this copy has been made by Kavirāja Umeśacandra Kaviratna in Chandernagor. H. P. Śāstrī later refers Julius Jolly to Cordier for help obtaining copies of this manuscript, although Cordier had already left Calcutta by then.

Although Cordier only ever wrote about H. P. Śāstrī’s medical manuscripts, he was evidently interested in non-medical ones as well — in his collection, there is a single photograph of the famous Kulālikāmnāya manuscript. It seems that he followed the same procedure as he did with the medical manuscripts — he created a transcription, based on the photograph, of a single folio of the Kulālikāmnāya.


The text of folio 26 verso of the Kulālikāmnāya manuscript was copied onto two leaves — it begins with three lines on one leaf (Sanscrit 1442.6, 4r) before switching to another one (Sanscrit 1442.3, 1v). The text is from the unpublished 7th chapter of the manuscript7 — a śabdarāśi enumerating the sounds of the Sanskrit syllabary and correlating each one with a part of the body and a particular manifestation of Śiva.8

Here, Cordier seems to be more interested in the form of the manuscript rather than in its text. Given H. P. Śāstrī’s extraordinary claim of its age, perhaps Cordier wanted a sample of its script and format for future reference. In contrast to the paper transcripts of medical manuscripts, Sanscrit 1442.3 is written on a palm leaf measuring 305 x 50 mm — exactly the same measurements that H. P. Śāstrī gives for his manuscript (12 x 2 inches). The script is also an imitation of its model, rather than being a transcription into modern Devanāgarī.9 There is a single stringhole in the same place as seen in van Kooij’s microfilm,10 and even the marginal title and folio number have been reproduced. However, the scribe — perhaps Cordier himself — stopped short of creating a perfect replica, seemingly unable to fit the text onto a single folio.11


van Kooij 1971, 882, detail. Reads "śrībhairav uvācaḥ |"

van Kooij 1971, 882, detail. Reads “śrībhairav uvācaḥ |” On this orthography, see Schoterman 1982, 18.

Sanscrit 1442.3 Ṣaṭsāhasrasaṃhitā, 1v, incipit. Courtesy of the BnF. Reads "śrībhairav uvācaḥ |"

Sanscrit 1442.3 Ṣaṭsāhasrasaṃhitā, 1v, incipit. Courtesy of the BnF. Reads “śrībhairav uvācaḥ |”

van Kooij 1971, 882 detail.

van Kooij 1971, 882 detail. H. P. Śāstrī believed that the shape of the numerals indicated the antiquity of the script.

Sanscrit 1442.3 1v, left margin.

Sanscrit 1442.3 Ṣaṭsāhasrasaṃhitā, 1v, left margin. Courtesy of the BnF.


On the recto side of Sanscrit 1442.3, Cordier has written a note based on information from H. P. Śāstrī:


Sanscrit 1442.3, 1r detail. Courtesy of the BnF.

Sanscrit 1442.3 Ṣaṭsāhasrasaṃhitā, 1r detail. Courtesy of the BnF.

Facsimilé d’un fol. du Kulālikāmnaya Tantra ; car. gupta.

(VII s. A.D. ?)

(Haraprasād Çāstrī ; Kathmandu, 1899.)

Cordier has put a skeptical question mark next to the suggested date; it seems that already in 1899, right after H. P. Śāstrī had returned from Nepal with the manuscript, Cordier was unconvinced by his claim. And he had good reason to be skeptical; at the end of 1898, Cordier had spent some time at Raghunātha Temple Library in Kashmir,12 so he would have had a lot of experience with Kashmiri manuscripts. Moreover, he knew Śāradā script very well; he had collated a manuscript of the Carakasaṃhitā in Śāradā script from Deccan College, in order to learn more about the Kashmiri recension of the text.


Sanscrit 1304 Carakasaṃhitā, inside back cover. Courtesy of the BnF. Cordier has copied the colophons from the two parts of the manuscript in Śāradā script.

Sanscrit 1304 Carakasaṃhitā, inside back cover. Courtesy of the BnF. Cordier has copied the colophons, in Śāradā script, from the two parts of the manuscript. In his note, he refers to the Deccan College manuscripts catalogue of 1888, which reproduces Bühler 1877.

In the margins of a printed edition of the Carakasaṃhitā, Cordier has noted variant readings from the manuscript, using Roman, Devanāgarī, and Śāradā script. At the end of the book, he has also affixed an additional page with a detailed description of the manuscript, including a study of the colophon and even a small sketch, showing the typical “portrait” orientation of a Kashmiri manuscript. Since he was fluent in both reading and writing Śāradā, the script of the Kulālikāmnāya — although an earlier form of Śāradā — must have been very familiar to him.


From hrön to hyperreal


If the Kulālikāmnāya that H. P. Śāstrī brought back from Nepal was unusual, and Sanscrit 1442.3 — Cordier’s attempt at reproducing one folio of it in palm leaf — is stranger still, then Sanscrit 1442.6 — which is a transcription of the same folio, containing three lines of text preceding Sanscrit 1442.3 — is utterly unique. Those lines are written not on palm leaf but on sāñcīpat — a folio made from agar bark, a tradition typical of Assamese manuscripts.


Cordier may have acquired these sāñcīpat folios as a curiosity; the manuscript contains 6 folios, all of which are blank except for the three lines written on one of them. Along with the bundle, Cordier has written a title card with a description of how these folios are made.


Sanscrit 1442.6, title card, recto. Courtesy of the BnF.

Sanscrit 1442.6 Ṣaṭsāhasrasaṃhitā, title card, recto. Courtesy of the BnF.

The text is a translation into French of a footnote to an 1894 article by Edward Albert Gait, which seems to have become the authoritative account of the production of sāñcīpat.13 Cordier has, additionally, translated Gait’s imperial measurements into metric units.

A tree is selected of about 15 or 16 years’ growth and 30 to 35 inches in girth, measured about 4 feet from the ground. From this the bark is removed in strips, from 6 to 18 feet long, and from 3 to 27 inches in breadth. These strips are rolled up separately with the inner or white part of the bark outwards, and the outer or green part inside, and are dried in the sun for several days. They are then rubbed by hand on a board, or some other hard substance, so as to facilitate the removal of the outer or scaly portion of the bark. After this, they are exposed to the dew for one night, and next morning the outer layer of the bark (নিকরি) is carefully removed, and the bark proper is cut into pieces of a convenient size, 9 to 27 inches long and 3 to 18 inches broad. These are put into cold water for about an hour and the alkali is extracted, after which the surface is scraped smooth with a knife. They are then dried in the sun for half an hour, and when perfectly dry are rubbed with a piece of burnt brick. A paste prepared from mātīmāh (Phaseolus aconitifolius) is next rubbed in, and the bark is dyed yellow by means of yellow arsenic. This is followed again by sun-drying, after which the strips are rubbed as smooth as marble.

Gait 1894, 109.

What was Cordier trying to do here, using sāñcīpat to copy out the Kulālikāmnāya? Was it simply a material that he had lying around, or was he trying to achieve a certain look? In his youth,14 Cordier had tried his hand at translating the Bower manuscript, “the most ancient Sanskrit medical treatise extant.”15 The Bower manuscript consists of birch bark folios — but in an oblong format mimicking palm leaves —, written in a script which A. F. Rudolf Hoernlé had dated to the early Gupta era. Had Cordier been trying to recreate this form — reasoning that the Kulālikāmnāya might be from the same time period —, but, lacking birch bark, used agar bark instead? Whatever the cause, the result — a manuscript commissioned or created by a French medical doctor in Bengal at the end of the 19th century — is a singular, original, and incredibly confounding written artefact: lines from a Śākta tantric text, written on sāñcīpat, reproducing an early Śāradā script copied from a photograph of a palm-leaf manuscript likely scribed by a Kashmiri in Nepal.


Hoernlé 1892, Plate 1. The Bower manuscript, part 1, leaf 1 verso.

Hoernlé 1892, plate 1. The Bower manuscript, part 1, leaf 1 verso.

Cordier was drafted at the beginning of the First World War and died shortly after in 1914, and this manuscript, along with the rest of his collection, was lost. It wasn’t until 1932 that Jean Filliozat chanced upon it in Paris chez un antiquaire, and, with the help of Sylvain Lévi, convinced the BnF to purchase the whole lot. It is due to this fortuitous chain of chance that the collection was kept together and finally conserved, so that we have enough context to piece together the story of Sanscrit 1442.6. Without knowing anything about it, what would a scholar have thought of it? What new subcategory of Eastern Nāgarī script might one have discovered in it, what fantastical conclusions might one have drawn about “Śākta theology” in Assam, not suspecting the labyrinth of increasingly unlikely coincidences that had led to the manuscript’s production?


Postscript: the Orientalist & the Śāstrī at his elbow


Inasmuch as these manuscripts are uncategorizable, so the characters involved in their production are equally hard to pin down. Palmyr Cordier was, primarily, a medical doctor. Yet, despite working full time at the hospital in Chandernagor, he managed to collect manuscripts from all over India, and he is now recognized as an eminent Sanskritist and Tibetanist. The eclectic conjunction of materials and circumstances that led to the creation of Sanscrit 1442.6 is due in large part to his voracious curiosity, which remains mostly undocumented. Although we know much about his work on Sanskrit medical texts, we have only hints about his forays into other Asian and even African languages;16 we know nothing about the non-medical Sanskrit manuscripts that he collected in India.

H. P. Śāstrī, too, is a singular historical figure. As an Indian Orientalist, he was uniquely positioned to build bridges between European and Indian scholars.17 In an inaugural address given to the Asiatic Society of Bengal in 1919, when H. P. Śāstrī was elected president, he said:

Oriental study is one of the great platforms in which east and west meet with mutual admiration and mutual sympathy. The east is proud of the results of the oriental studies, because they belong to the east; the west is proud because they have given a new significance to these studies, and  their results. The west is grateful to the east for revealing a great civilization that has passed away and in which they find so much to study and to reflect, and the east is grateful that they have got a new light from their old things. I wish this enthusiasm to last long and bring about the desired end.18

His enthusiasm did not last long. Eight years later, in a speech given at the Fifth Indian Oriental Conference, his view on European Orientalists had changed dramatically:

At the present moment there is a large body of men who go as Sanskrit scholars without knowing a letter of Sanskrit. There are others again who tax the brains of poor Śāstris and make big name as Oriental scholars. At the conference of Orientalists held under the Presidency of Sir Harcourt Butler in 191119 a very great man told the august assembly that without two Śāstris at their elbows they cannot be Oriental scholars. Such Oriental scholarship should be discouraged. The Śāstris should be trained for Oriental scholarship. A historical sense should be awakened in their minds.

His change of heart was undoubtedly a result of the Indian independence movement that was gaining traction at the time, but, perhaps, he was also reflecting on his own long career as, very literally, a Śāstrī to many European Orientalists like Cecil Bendall. Yet, in this passage, he does not seem to be describing himself. Unlike the stereotype he paints,20 and unlike the many — often anonymous — Indian paṇḍits who worked under him, he was able to move fluently between East and West, being well-respected in both worlds. And it is due to this respect that his early assessment on the date of the Kulālikāmnāya manuscript was cited again and again, without question. But as it turns out, manuscripts, like people, are not so easy to classify.


Suggested citation

Li, Charles. 2022. “Kathmandu, Calcutta, Paris: the life of a Tantric manuscript.” Texts Surrounding Texts: Satellite Stanzas, Prefaces and Colophons in South-Indian Manuscripts (Paris BnF and Hamburg Stabi collections). 1 March. https://tst.hypotheses.org/3191

Charles Li

Charles is the resident postdoctoral researcher at the TST project in Paris.

More Posts - Website

  1. This is not to diminish his other work, such as his Bengali science fiction novella serialized in Baṅgadarśan and later translated into English in as The Triumph of Valmiki. Sylvain Lévi wrote a review of it for the Journal asiatique, which does not seem to have been published. However, in a long article for the Revue de Paris, he mentions it in passing as a cross between Paradise Lost and the Rāmāyaṇa. []
  2. Bendall interpreted the date as A.D. 859 or A.D. 857, but this was revised by later scholars to 828 C.E. (Goodall 1998, xliii, Sanderson 2001, 5). For more on the Cambridge collection, see Formigatti 2017. []
  3. See Schoterman 1982, 16. []
  4. Witzel 1996. Also available via academia.edu. []
  5. The photographs are presumably at the BnF, although they have not been located. []
  6. See Wujastyk 1998, 207. []
  7. H. P. Śāstrī records chapter 7 spanning folios 24 to 36. See also Schoterman 1982, 218ff. []
  8. See also Padoux 1963, 244, Goudriaan & Schoterman 1988, 442, Sanderson 1990, 46, etc. []
  9. This practice of copying an archaic script is not limited to modernity or to scholars; inscriptions also seem to mimic archaic characters. []
  10. Witzel, describing a diferent manuscript, suggests that this stringhole placement points “to a fairly early date” (1996). []
  11. See Sanscrit 1442.6 below. Sanscrit 1442.3 has six lines, whereas H. P. Śāstrī reports 7 lines per page. []
  12. Roșu 1990, xciv. []
  13. See Datta 1970, 110 & 136Das 2006, 83ff, Goswamee 2006, 73ff, & Gupta 2010, 51ff. All four seem to quote Gait, explicitly or implicitly. []
  14. Roșu 1990, lxviii. []
  15. From an 1895 article in the British Medical Journal. []
  16. Roșu 1990, lxviii. []
  17. As well as Japanese scholars. He was also instrumental in organizing a donation of manuscripts from the Mahārāja of Nepal to the Bodleian Library. []
  18. This is particularly interesting when paired with his 1899 History of India, which ends with an ode to British rule. []
  19. He seems to be referring to The Conference of Orientalists including Museums and Archæology Conference held at Simla, July 1911. []
  20. In fact, Wladimir Ivanow, who knew and admired H. P. Śāstrī, had complained about a similar instance of “parasitic plagiarism”. []

OpenEdition suggests that you cite this post as follows:
Charles Li (March 2, 2022). Kathmandu, Calcutta, Paris: the life of a Tantric manuscript. Texts Surrounding Texts. Retrieved July 18, 2024 from https://doi.org/10.58079/uxzv

Charles Li

Charles is the resident postdoctoral researcher at the TST project in Paris.

You may also like...

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Search OpenEdition Search

You will be redirected to OpenEdition Search