India's Traditional Knowledge Digital Library Has Transcribed 519,978 Medicine and Yoga Formulations Into Five Languages, and 375 Patent Applications Have Been Refused, Withdrawn, Amended or Set Aside on Its Evidence Since 2009

The Traditional Knowledge Digital Library, usually called TKDL, holds 519,978 formulations and techniques from India’s medical systems, transcribed out of Sanskrit, Hindi, Tamil, Malayalam, Arabic, Persian, Urdu, Bhoti and other languages into English, French, German, Spanish and Japanese. It was set up in 2001 jointly by the Council of Scientific and Industrial Research (CSIR) and the then Department of Indian Systems of Medicine and Homeopathy, now the Ministry of Ayush. Eighteen patent offices have signed agreements that let their examiners search it, and a written reply in the Lok Sabha on 13 March 2026 reported that since 2009 a total of 375 patent applications have been refused, withdrawn or deemed withdrawn, amended or set aside on TKDL evidence. The Press Information Bureau (PIB) published the reply on 17 March 2026, and those are the most recent official figures available as of 6 November 2026.

What the library holds

PIB’s breakdown of the 519,978 entries is 148,846 from Ayurveda, 265,496 from Unani, 89,403 from Siddha, 9,097 from Sowa Rigpa (the Tibetan medical tradition of Ladakh and the Himalaya) and 7,136 yoga practices. The same release counts 2,935 surgical methods, devices, therapies and dietary interventions among them. Earlier counts were much lower: V. K. Gupta’s WIPO Magazine article, written when the database was newer, puts it at about 226,000 formulations, so only the March 2026 count is used here. The medical tradition behind these texts is covered in Charaka and the ancient textbook of medicine and the Sushruta Samhita, and the yoga entries connect to the International Day of Yoga.

CSIR’s own TKDL page describes the method. Each entry is rewritten in the format a patent examiner expects, and a purpose-built scheme called the Traditional Knowledge Resource Classification sorts the material into several thousand subgroups. The library draws only on texts in the open domain, which anyone in India or abroad could obtain. CSIR says it works as a bridge between those texts, which count as prior art (earlier published knowledge that can defeat a claim to novelty), and the examiners who need to see them.

The turmeric patent of 1995

The case that started the project concerned a kitchen spice. The US Patent and Trademark Office granted patent 5,401,504 on 28 March 1995 to Suman K. Das and Hari Har Cohly of the University of Mississippi Medical Center, for a method of promoting wound healing by giving turmeric by mouth or on the skin, according to Down To Earth’s 1997 reporting. Business Standard reported on 25 August 1997 that the revocation order was passed on 13 August 1997 and that CSIR’s director general had announced it days earlier, and Down To Earth gives the same 13 August date for the withdrawal. Nature carried the news on 4 September 1997 under the headline “US patent office withdraws patent on Indian herb”.

Raghunath Mashelkar, then CSIR’s director general and the author of a 2022 account of the case, writes that CSIR filed its request for re-examination in October 1996 and cited 32 references, ranging from centuries-old Sanskrit texts to material in Hindi, Urdu and other Indian languages. He dates an order rejecting all six claims to March 1997, five months after the request. The sources therefore differ on whether the rejection came in March or August 1997, and each is named here.

Down To Earth’s contemporary report adds a detail about the evidence. It says the evidence CSIR put before the US office was a paper published in 1953 in the Journal of the Indian Medical Association by M. L. Gujral, N. K. Chowdhury and P. N. Saxena, and that Ayurvedic scriptures could not serve as conclusive evidence under the US rule it describes: foreign knowledge counted as prior art only when it had been patented or described in a printed publication. Mashelkar’s later account stresses the Sanskrit references instead, so the two accounts differ on how much of the evidence came from the older texts. Both agree that the case turned on documents the examiner could find and cite, which is the gap a library of transcribed texts is built to fill.

CSIR spent Rs 5 lakh (about 500,000 rupees) in legal fees and other charges on the challenge, Business Standard reported, and quoted Mashelkar as saying it appeared to be the first time a US patent based on a developing country’s traditional knowledge had been successfully challenged at the US office.

The neem patent in Europe

A second case ran in parallel at the European Patent Office (EPO). European patent 0436257, for controlling fungi on plants with a neem oil formulation, was granted in September 1994 to the US Department of Agriculture and the company W. R. Grace, according to the Financial Express report of 10 March 2005. The opposition filed in June 1995 came from Vandana Shiva’s Research Foundation for Science, Technology and Ecology in Delhi, the International Federation of Organic Agriculture Movements and Magda Aelvoet, then a Green member of the European Parliament. The opposition was lodged by those three and not by CSIR, and the TKDL did not yet exist.

The EPO’s own decision record shows how the case unfolded. The opposition division took the testimony of a witness, A. D. Phadke, at oral proceedings on 9 and 10 May 2000, and it considered proven that a hydrophobic neem oil extract had been used on crops in the Pune and Sangli districts of Maharashtra in 1985 and 1986 against pests and fungal diseases. It posted its written decision revoking the patent on 13 February 2001. The patent holders appealed, and on 8 March 2005 the Technical Board of Appeal dismissed the appeal. The Board decided the case on a 1981 scientific paper by H. B. Singh and U. P. Singh in Australian Plant Pathology and left open whether the Indian prior use was proven.

Why examiners missed Indian medicine

Mashelkar’s account of what followed is that, during a visit to the US patent office in Washington, he asked why its search had not turned up the Indian prior knowledge. He writes that the answer was the International Patent Classification, the filing scheme every examiner searches by, which had no place for traditional medicine. The WIPO Magazine article by V. K. Gupta, the TKDL’s architect, describes the same gap: until 2005 only one subgroup, A61K35/78, covered medicinal plants.

Gupta also records that around 2001 a TKDL expert group estimated that some 2,000 patents a year touching Indian medicinal systems were being wrongly granted worldwide, because the relevant texts existed only in languages examiners could not read. That is an estimate, and WIPO material on the TKDL notes that not every patent built on Indian systems is wrongly granted. India took the classification problem to WIPO. Gupta and Mashelkar both describe a task force of China, the European Union, India, Japan and the United States, and CSIR says the creation of the TKDL and its classification system led to about 200 subgroups under A61K 36/00 in the International Patent Classification. Gupta writes that in 2004 it was agreed to link the 27,000 subgroups of India’s own scheme to the international one.

How it works in a patent office

Access is granted under a non-disclosure agreement approved by India’s Cabinet Committee on Economic Affairs, according to CSIR. Examiners may use the database for search and examination, and may reveal its contents to a third party only to cite it. CSIR’s page lists seventeen offices: the European Patent Office, the United States, Japan, the United Kingdom, Canada, Germany, Australia, India, Chile, Malaysia, Russia, Peru, Spain, Denmark, France, the Eurasian Patent Organization and the Philippines. PIB’s 2026 count is eighteen, so either one office has been added since CSIR’s list was written or the two count differently.

The CSIR unit also watches published applications and files third-party observations and pre-grant oppositions, a step that lets anyone submit prior art before a patent is granted. Gupta compares the costs. He writes that opposing a granted patent takes five to seven years and between 0.2 and 0.6 million US dollars on average, while an objection supported by TKDL evidence at the pre-grant stage takes 3 to 20 weeks. His early results at the EPO were 215 applications flagged since July 2009, two where the office reversed its intention to grant, one where the applicant narrowed its claims and 33 withdrawals.

The running totals have changed with each report. Mashelkar wrote in 2022 that 265 applications had been withdrawn, amended or set aside, and that 14 patent offices had access. CSIR’s page says close to 370, and the March 2026 PIB release says 375 across 18 offices. They are different dates and slightly different wording, and the 375 is the figure in the Lok Sabha reply.

What the law and WIPO now say

India’s Patents Act, as amended in 2005 and reproduced in WIPO’s database of traditional knowledge laws, excludes from patents an invention that in effect is traditional knowledge (section 3(p)). It also lets anyone oppose a granted patent within a year on the ground that the claimed invention was anticipated by knowledge, oral or otherwise, available within a local or indigenous community in India or elsewhere (section 25(2)(k)). PIB adds that the Controller General of Patents, Designs and Trade Marks has issued guidelines for examining Ayush-related inventions, drawn up with the Ministry of Ayush, and that the Biological Diversity Act, 2002 regulates access to biological resources and associated traditional knowledge through the National Biodiversity Authority. How Ayurvedic claims fare in clinical trials is a separate question, covered in the science of Ayurvedic claims.

At WIPO, member states adopted a Treaty on Intellectual Property, Genetic Resources and Associated Traditional Knowledge on 24 May 2024. WIPO’s summary says it requires patent applicants to disclose the country of origin of genetic resources, or the Indigenous or local community that supplied associated traditional knowledge, and that it enters into force three months after 15 ratifications or accessions. CSIR’s page says the treaty’s Article 6 recommends information systems such as databases of associated traditional knowledge, which it calls a recognition of the TKDL’s approach.

Gaps the project is still closing

The TKDL covers published texts of the Indian systems of medicine and yoga. Down To Earth’s 1997 report quoted the campaigner Suman Sahai on farmers’ knowledge of neem as a pesticide, which was passed on by word of mouth and had no documents behind it. CSIR’s page says digitisation of traditional agricultural practices and of traditional metallurgical knowledge has been started.

For basmati rice, a case outside the TKDL’s medical scope, Mashelkar writes that India filed a re-examination request in April 2000 against RiceTec’s US patent and that RiceTec later withdrew the crucial claims. CSIR’s page says the Agricultural and Processed Food Products Export Development Authority spent about seven crore rupees (70 million rupees) on legal fees to get a few of those basmati claims revoked.