Diabetes mellitus is a biomedical diagnosis that rests on blood glucose measurement. A healer without a glucometer recognises something else: a cluster of signs such as frequent urination, thirst, loss of weight, wounds that heal slowly, or urine that attracts ants. Hindi and regional terms such as madhumeh and the borrowed word "sugar" circulate alongside older local categories. A survey that simply asks for "plants for diabetes" therefore records a mixture of answers to different questions. The distinction between the informant's own category and the researcher's is basic to field method (Martin, 1995), and for this topic it decides whether the results mean anything.

Specialist knowledge is unevenly held. Village herbalists, known by titles that vary with community and district, treat patients from several settlements and often guard their formulas. General household surveys underrepresent what they know, while a study confined to two or three famous healers cannot show which uses are shared. The design proposed here follows the healers' consensus approach, in which agreement among independent specialists is taken as a sign that a use is culturally established and worth closer study (Trotter and Logan, 1986; Heinrich et al., 1998).

India's Biological Diversity Act regulates access to biological resources and associated knowledge through the National Biodiversity Authority, State Biodiversity Boards and local Biodiversity Management Committees (Government of India, 2003). It gives domestic effect to the principles of prior informed consent and benefit sharing later elaborated in the Nagoya Protocol (Secretariat of the CBD, 2011). The Code of Ethics of the International Society of Ethnobiology adds obligations that law does not cover, among them educated prior informed consent, confidentiality where requested, and the return of results in usable form (International Society of Ethnobiology, 2006). These are treated below as design constraints, not as paperwork to be completed afterwards.

Work begins with visits to the village assembly, the Biodiversity Management Committee where one exists, and the forest and health offices of the district. The purpose of the survey, what will and will not be published, and what the community receives in return are discussed before any interview. Healers decide individually whether to take part and whether particular formulas are to remain unpublished. Consent is sought in the local language, orally where literacy is limited, and recorded. Review by an institutional ethics committee and any approval required under the Biological Diversity Act are obtained at this stage.

A short preliminary round of open conversations with a few healers and health workers establishes how the condition is named, which signs are taken to indicate it, and whether healers treat patients who have already been diagnosed at a clinic. From this a glossary and two or three screening questions are prepared. In the main interviews each remedy is then tagged by the basis on which the healer assigns it: a biomedical diagnosis reported by the patient, the healer's own recognition of signs, or a related complaint such as excessive urination or non-healing sores.

Blocks are chosen to span the main tribal communities and forest types of the study area. Within each, healers are identified by asking village leaders, health workers and other healers, and the chain of referral is written down. This is purposive sampling and is reported as such; it supports statements about what specialists know, not estimates for the population. Recruitment continues in each block until further referrals lead mainly to people already interviewed. Sex, age, community, years of practice and the way the healer learned are recorded for every participant.

Each healer is interviewed alone with a semi-structured guide: plants used, local names, part, preparation, quantity as the healer measures it, duration, accompanying dietary advice, perceived effects and unwanted effects, and where and when the plant is gathered. Free listing at the start gives an unprompted inventory. The interview is followed by a walk to the places where the healer collects, which removes much of the ambiguity of names given indoors and allows plant material to be gathered in the healer's presence (Alexiades, 1996). Mixtures are recorded as mixtures, with every ingredient, since many remedies are polyherbal.

Every plant named is collected, where abundance permits, as a fertile specimen with a collection number that links it to the interview. Specimens are pressed, dried and mounted by standard herbarium practice (Bridson and Forman, 1998), identified against regional floras and reference collections, and deposited in a recognised public herbarium, with duplicates at a local institution. Scientific names are given with author citation and family and checked against a current global checklist. Unverifiable names are the single commonest reason why field data cannot be reused (Rivera et al., 2014). Material sold in markets or bought as dried drug is vouchered separately and labelled as such.

The unit of analysis is the use report: one healer citing one species for one purpose. Interview notes are converted to a table with one row per use report and columns for species, voucher, part, preparation, route, the basis of assignment from Step 2, and whether the plant is used alone or in a mixture. A written codebook defines each category. Two people code a portion of the interviews independently and settle differences before the remainder is coded, so that the table can be audited against the notes.

Agreement is summarised with a small number of indices whose formulas are stated. The informant consensus factor, (Nur − Nt)/(Nur − 1), relates the number of use reports in a category to the number of taxa cited (Heinrich et al., 1998). The fidelity level expresses the share of a species' citing informants who name it for the category of interest (Friedman et al., 1986). Relative frequency of citation shows how widely a species is known (Tardío and Pardo-de-Santayana, 2008). The underlying use-report table is published with the indices. Each species is then set against the pharmacological and clinical literature, noting whether the reported use is already documented and whether any toxicity is known.