The Bower Manuscript, Dug Up Near Kucha, Preserves Sanskrit Medical Recipes in Gupta Script on Birch Bark, Among Them a Formulary Called the Navanitaka

The Bower Manuscript is a bundle of Sanskrit texts on birch bark, written in the Gupta-period script of northern India and found in a ruined Buddhist stupa near Kucha, an oasis town on the northern Silk Road in what is now Xinjiang, China. Three of its seven parts are medicine. Longest is the Navanitaka, a formulary of recipes that its writer planned in sixteen chapters and that opens with a bow to the Buddhas. Oxford’s Bodleian Library holds the leaves as MS. Sansk. c. 17.

A. F. Rudolf Hoernle, principal of the Calcutta Madrasah, edited the leaves for the Archaeological Survey of India in a series that ran from 1893 to 1912. After about a week with the manuscript in April 1891 he wrote that it was the oldest Indian written book then known to exist. That was a claim of 1891, made before later finds in Central Asia, and the sources used here date the leaves anywhere from the 4th to the 6th century CE.

A treasure-seeker, a lieutenant and a stupa

The manuscript is named after Lieutenant Hamilton Bower, who was at Kucha (spelled Kuchar in his day) on a confidential mission for the Government of India, in search of the murderer of a man named Dalgleish. A letter Bower wrote to Hoernle on 17 August 1892 tells the story. A man in Kuchar told Bower of an underground city and said he had gone there to dig for treasure but had found only what he called a book. Bower asked to see it, and the man came back with the manuscript. Bower was glad to buy it for a small sum.

Bower sent it to Colonel J. Waterhouse, president of the Asiatic Society of Bengal, in September 1890, and it was shown to the Society on 5 November 1890. Attempts after the meeting to read it failed. Hoernle, on leave in Europe, saw a report in the Bombay Gazette at Aden on his way back. He took the manuscript over from Waterhouse in Calcutta early in 1891 and published the first decipherment in the Society’s Proceedings for April 1891.

Found in 1889 or in 1890

Sources do not give the year of the find the same way. The Encyclopaedia Iranica entry on Hoernle, by Ursula Sims-Williams, says the manuscript was discovered near Kucha in 1889. A Wire Science write-up of the Scrolls & Leaves podcast (December 2021) likewise has Bower tracking the killer in 1889, though it does not say when the book was found. Hoernle’s introduction puts the manuscript in Bower’s hands early in 1890.

That date comes from Bower’s travel account in the Geographical Journal of 1895. Bower left Shahyar on 6 March on his way back to Kashgar, which he reached on 1 April. Counting backwards through the visit to the find-site, Hoernle concluded that Bower received the manuscript on 2 or 3 March 1890, and that the Kuchar men had dug into the stupa a few days earlier, in February 1890. His introduction adds that the French expedition of Paul Pelliot, which had explored the Kuchar district, might be expected to settle any remaining doubt about the exact place and time.

Opinion on the place was also divided. Georg Bühler first announced the discovery as coming from the ruins of the underground city of Ming-oi near Kuchar, and Hoernle points out that Bower said only that it came from a stupa just outside the city. A local tradition in Kuchar, on the other hand, linked the Bower leaves to the larger Qutluq Urda stupa nearer the town, from which other collectors’ manuscripts came. Rejecting both, Hoernle concludes that it was a stupa on the ridge above the caves of Qum Tura.

Seven texts in one bundle

The bundle, Hoernle found, holds seven parts in the hands of four scribes. Counts of the surviving leaves differ: the folio counts he gives for the seven parts (five, thirty, four, five, six, four and two) add up to 56, while an online summary gives 51. Parts I to III are medical. Parts IV and V are tracts of divination, and the discussion in his introduction centres on oracles worked by throws of dice. Parts VI and VII are two copies of one charm against snake bite, the Mahamayuri, which he describes as a charm of great repute among Buddhists. His 1891 note gave a tentative translation of one set of three divination sayings, of which the first begins “Thy intelligence is spoiled” and the second promises quick deliverance from diseases.

In Hoernle’s reading the whole bundle belonged to one Buddhist monk, whose name appears to have been Yasomitra, and was kept in a monastery near Kuchar. Finding the leaves in a stupa’s relic chamber, he took it that the owner was held in high regard. Two of the four scribes, he judged, were Indian emigrants. Birch that yields writing bark does not appear to grow in Eastern Turkestan, and most of the manuscript is on poor, damaged bark, which Hoernle took to be the remains of a stock the emigrants brought from India.

Garlic and ten sages

Part I is five leaves and breaks off. It opens with a verse description of a Himalayan mountain where ten sages live: Atreya, Harita, Parasara, Bhela, Garga, Sambavya, Susruta, Vasishtha, Karala and Kapya. In the translation they roam hundreds of times in company, asking about the tastes, properties, forms and names of medicinal plants.

Susruta sees a plant with white bulbs and goes to ask the sage King of Kasi what it is. Hoernle reads the King of Kasi as most probably Divodasa, also known as Dhanvantari. His reply names garlic. According to the text, when the lord of the Asuras drank the churned nectar, Vishnu cut off his head, and drops from the severed windpipe fell on the ground and became garlic. For that reason, the passage adds, Brahmins do not eat it. A note by Hoernle lists the Ashtanga Hridaya as giving a long account of garlic that substantially agrees with this one, and as telling the usual version, in which the Asura is Rahu. The garlic tract itself has 43 verses in eighteen or nineteen different metres, and goes on to the plant’s properties and recipes.

The Navanitaka

Part II, the Navanitaka, was written on 33 numbered leaves, of which 30 survive. Its introduction begins “Salutation to the Tathagatas” and promises, in Hoernle’s translation, a standard manual of “the foremost formulas of the Maharshis” made up by them of old. Navanitaka means cream, and Hoernle’s note takes the name to mean a collection of approved extracts from the older medical writers.

Its sixteen announced chapters run through powders, medicated ghee, oils, mixed remedies for various diseases, enemas, elixirs, gruels, aphrodisiacs, eye-salves and hair dyes, then three chapters on single substances (the chebulic myrobalan fruit, bitumen and the plumbago root), and then children, barren women and mothers. Chapters fifteen and sixteen are missing, and the text breaks off in the fourteenth. The introduction closes with a rule that the book is not to be given to a man with no son, no brother or no disciple.

First comes the Talisa powder: talisa leaves (Hoernle’s identification is Taxus baccata, the yew, which he marks as uncertain), black pepper, dry ginger and long pepper in rising proportions, with cinnamon, cardamom and sugar, for cough, asthma and loss of appetite. A note by Hoernle finds the same formula in the Charaka Samhita and two other Sanskrit compendia, all three adding a verse that allows the powder to be boiled with sugar and made into pills. In his 1909 paper for the Royal Asiatic Society he went further and argued that the Navanitaka’s formulas correspond to passages in the Charaka Samhita and that in some cases the Navanitaka seems to preserve the older text, a claim that rests on his reading of the variants. For the textbook itself see the Charaka Samhita article.

How old is the handwriting

Everyone who dates the leaves uses the shape of the letters, and Hoernle changed his mind more than once. In April 1891 he wrote that the manuscript was not later than the end of the 5th century. He argued from the letter ya, whose middle form appears in dated inscriptions only between about 470 and 530 CE, and from the absence of a place-value zero, which he placed within the 6th century. He gave about 500 CE.

His later essay put the date at the middle of the 5th century. By 1909 he wrote in the Royal Asiatic Society’s journal that it belonged about a century earlier, and in his introduction he says good reasons had appeared for moving it back by about three-quarters of a century, with the stage of the letter ya seen in the leaves belonging to the second half of the 4th century. Google Arts & Culture, in a record supplied by the National Centre for Biological Sciences, lists the manuscript as 4th to 6th century. Jeremy Norman’s page gives the same range and adds that later palaeographic work favours about the first half of the 6th century, without naming the scholars. None of the sources used here reports a radiocarbon test of the Bower leaves, unlike the Bakhshali birch-bark manuscript, whose 2017 date was revised in 2024.

Its script is the northern Gupta hand, the writing of the empire covered in the Gupta age article. Hoernle noted that the leaves write numbers with separate signs for tens and hundreds, not by place value. For the zero’s own history see India Gave the World Zero.

What came after

Iranica says the interest Hoernle’s report aroused led the great powers into a competitive search for manuscripts and antiquities in Central Asia. His introduction ties it to concrete steps: in 1893 the Government of India told its political agents in Kashmir, Ladakh and Kashgar to look out for ancient manuscripts, and he traces the first expedition of Aurel Stein into Chinese Turkestan, in 1900 to 1901, to the stimulus of the Bower discovery. Until the text was complete the leaves stayed with Hoernle, who returned them to Colonel Bower in April 1898.