Sources

bodhi cites and reproduces images from open-access museum collections and named scholarly publications. The editorial floor is the same on every article: at least four sources, drawn from at least two institutions, with every image traceable to its catalogue page. No piece rests on a single collection, a single scholar, or a single primary text.

This page lays out the full sourcing strategy — what counts toward the floor, which collections feed which tier, how images are verified, where the unavoidable gaps are, and what bodhi explicitly does not cite.

The four-source floor

Every published article carries four sources of distinct kinds. The distinctions are load-bearing — a piece with four museum accession lines and no scholar fails the floor; so does a piece with four scholarly references and no primary text.

A Tier-1 or Tier-2 museum accession — a named, accessioned object in an open-access collection, used either as the hero work or as a cited comparand. The accession number, period, and rights line are recorded in the article frontmatter, not the prose.
A named-scholar passage with a page reference — a citable position from a peer-reviewed monograph, museum catalogue, or refereed journal article. “See Bogel (2009)” is not a citation; “Bogel 2009, pp. 142–146” is. The reading pool below names the working scholars per cluster.
A primary-text citation — a sūtra, ritual manual, mandala text, or pre-modern catalogue, named with its canonical title in the source language and (where available) its Taishō or equivalent number. The English translation cited is named and edition-pinned.
A cross-institutional reference — at least one of the sources above resolves to a different institution from the hero image’s holding collection. The aim is that every iconographic claim is anchored to work in two or more collections, not one.

The four-source floor is the minimum. Iconographically dense pieces (disambiguation tables, mandala readings, sculptural-lineage articles) typically carry six to twelve sources.

Image sourcing — the three tiers

Tier 1 — CC0 and equivalent public-domain dedications

The image floor. Cannot be retroactively revoked. The hero image of every article is Tier 1, and at least one secondary work cited is Tier 1.

The Metropolitan Museum of Art — Open Access since 2017, CC0 for records flagged isPublicDomain: true. ~492,000 OA images; Asian Art ~35,000 objects. Strong on Japanese Buddhist sculpture, Chan/Zen painting, Edo-period zenga (Hakuin, Sengai). metmuseum.org
Cleveland Museum of Art — Open Access since 2019, CC0. ~30,000+ images. Connoisseur’s first stop for Chan figure painting; major holdings in Goryeo Buddhist painting and Japanese Nanbokuchō ink scrolls. clevelandart.org
Smithsonian National Museum of Asian Art (Freer & Sackler) — CC0 via the Smithsonian Open Access programme since 2020. Deepest US East Asian holding outside the Met; the Freer’s photography programme is best-in-field. asia.si.edu
Art Institute of Chicago — CC0 for PD works. Strong Japanese woodblock prints (Buckingham), Buddhist sculpture, ink painting. Cleanest API in the field. artic.edu
Harvard Art Museums — mixed; CC0-equivalent OA for records with imagepermissionlevel: 0. Strong Chinese Buddhist sculpture, Japanese painting. harvardartmuseums.org
Los Angeles County Museum of Art — ~20,000 CC0 images, one of the earliest US museums to adopt open access (2013). World-class Himalayan-Esoteric holdings relevant to Shingon roots. lacma.org
Minneapolis Institute of Art — open-access subset, CC0. new.artsmia.org
National Gallery of Art, Washington — open-access, CC0. nga.gov
Princeton University Art Museum — many PD works downloadable; excellent Chinese painting (Edward Elliott Family Collection of literati ink painting, much Chan-adjacent). artmuseum.princeton.edu
Rijksmuseum (Rijksstudio) — CC0 for ~700,000 OA images. Smaller East Asian holding but excellent Edo Japanese painting and prints. rijksmuseum.nl
Musée Cernuschi (Paris Musées) — CC0 since 2020. Chinese bronzes, Buddhist sculpture, Japanese painting; genuinely useful and under-known. parismuseescollections.paris.fr
National Diet Library, Japan — pre-1947 holdings, public domain; IIIF Presentation API across much of the collection. Pre-modern Japanese Buddhist manuscripts, illustrated sūtra, Edo-period Zen woodblock printed books. ndl.go.jp
Wikimedia Commons — CC0, CC-BY, CC-BY-SA, PD-art (per-file check mandatory). The only realistic way to get site photography of in-situ sculpture and architecture at Longmen, Yungang, Nara temples (Tōdai-ji, Kōfuku-ji), Kamakura. commons.wikimedia.org
Library of Congress — most holdings “no known restrictions”; pre-1928 missionary-era Chinese temple photography. loc.gov
Internet Archive — pre-1929 scholarly PD plates (Sirén Chinese Sculpture 1925, Soper, Fenollosa, Pal catalogues). Often the only PD source for objects in private collections or destroyed sites. archive.org

Corpus target: at least 75% of all reproduced images Tier 1.

Tier 2 — open-access in practice

Broad re-use granted under terms that differ from CC0, sometimes materially. Re-verified annually against the official wording.

ColBase — Independent Administrative Institution, National Institutes for Cultural Heritage, Japan. The single most important source for Japanese Buddhist art globally. The unified portal indexes Tokyo, Kyoto, Nara, and Kyushu National Museums. colbase.nich.go.jp
e-Museum — National Institutes for Cultural Heritage. National Treasures and Important Cultural Properties under the same terms as ColBase. emuseum.nich.go.jp
National Museum of Korea — Korea Open Government License (KOGL) Type 1 for most images: free including commercial use, attribution required. The only major source for Korean Buddhist art at scale (Three Kingdoms, Unified Silla, Goryeo). museum.go.kr
International Dunhuang Project (BL hub, with mirrors in China, Russia, Japan) — per-institution terms; BL’s IDP images generally under generous open licences. Largest body of Tang-era Buddhist material — manuscripts, silk paintings, banners, sūtra fragments dispersed by the Stein, Pelliot, Otani, and Oldenburg expeditions. idp.bl.uk

ColBase grants free, application-free use for commercial and non-commercial purposes provided the source is clearly indicated. The licence text, verbatim:

「出典を明記いただければ、申請不要、かつ無料で、営利・非営利問わずご利用いただけます」

Every ColBase-sourced image on bodhi carries the verbatim attribution line in its caption block:

出典：ColBase（https://colbase.nich.go.jp/）

The Japanese attribution is preserved as required by the source — not paraphrased, not translated. ColBase terms are re-verified annually against the official wording.

Tier 3 — text-only, never reproduced

Canonical works under restrictive licences. bodhi cites their accession records and links to their catalogue pages; no Tier-3 images are reproduced on the site.

The British Museum — CC BY-NC-SA. Non-commercial clause is incompatible with bodhi’s editorial licence (CC BY-SA 4.0).
Victoria and Albert Museum — CC BY-NC. Same reason.
Palace Museum, Beijing — all rights reserved. ~1.8 million objects including the imperial Buddhist collection; no open access.
Dunhuang Academy (e-dunhuang.com) — restricted. IDP is the legally usable channel for portable Dunhuang material.
Shōsō-in (Imperial Household Agency) — not open. Annual exhibition catalogues are the reproduction channel; older PD-by-age catalogues exist on Internet Archive.
Major temple collections — Daitoku-ji, Tōfuku-ji, Tō-ji, Kōzan-ji, Hōryū-ji, Kōfuku-ji and others — almost no temple has released its collection under open licences. Formal permissions only.
Auction houses — Sotheby’s, Christie’s, Bonhams — provenance citation only; bodhi does not reproduce auction-house photography.
Musée Guimet (Paris) — most images sold through RMN-Grand Palais. Functionally inaccessible despite the collection’s importance.

For Tier-3 works, the article cites the accession number and links to the holding institution’s own catalogue page. The reader sees no image; the citation chain is intact.

The diversification cap

To prevent single-source dependency, no single institution provides more than 35% of bodhi’s image heroes. The Met and ColBase will be the largest contributors; the cap stops a hypothetical policy change at one collection from forcing widespread takedowns. The cap is enforced in the publishing pipeline against the running cleared-image manifest, not post-hoc.

Verification chain

Every wired image is verified at the file-page level — not the search-results page, not the API summary. The production build is gated by check:uncleared: any image without a verified rights line under /public/images/cleared/ fails the build.

The chain, in order:

The image is downloaded from the source collection’s official URL (museum file page, IIIF manifest, or Wikimedia Commons mirror).
The Commons SHA1 (when applicable) and the museum’s file-page rights statement are recorded.
The image, rights line, and source URL travel together as a single WorkImage block in the article frontmatter — bound at write time, never reconstructed downstream.
ColBase, KOGL, IDP-partner terms are re-verified annually against the official wording.

The image-rights detail is on the dedicated Image rights page, including the per-tier attribution structure and the takedown procedure.

Visual triage

Search APIs sometimes return images that do not match the requested accession. bodhi inspects every candidate image visually before wiring, matching figure-specific iconographic markers (eleven heads stacked for Jūichimen, sword and lasso for Fudō, descending procession for raigō, the radial fan of eighteen arms for Juntei, and so on).

A wrong image is worse than no image. If no Tier-1 or verified Tier-2 image survives this check, the article ships without inline reproduction; the prose stands, and the reader is linked to the collection’s own catalogue page.

The China-mainland gap

Palace Museum Beijing, Dunhuang Academy, Shanghai Museum, and the provincial museums hold the largest body of Chinese Buddhist art on Earth. None release open-licensed images at usable resolution. There is no Open Access movement comparable to the West.

bodhi covers Chinese Buddhist art through five routes:

Diaspora collections — for nearly any Chinese Buddhist subject, the Met, Smithsonian, Cleveland, or Princeton holds a comparable CC0 object. Editorial scope is built around what is available open-licensed, not around what is “definitively best.”
Wikimedia in-situ photography — Longmen, Yungang, Maijishan, exterior Mogao extensively photographed by visiting Wikimedians under freedom-of-panorama (see below).
IDP for portable Dunhuang material — manuscripts and silk paintings dispersed by Stein, Pelliot, Otani, and Oldenburg expeditions split among the British Library, BnF, the Hermitage, and Ryukoku University.
Pre-1930 publication plates via Internet Archive — Sirén photographed Longmen and Yungang in 1925 before mid-century vandalism; the plates are public domain and remain the only PD reproduction for many lost or damaged figures.
Wikidata-driven sourcing — Wikidata Q-IDs resolve to whichever collection holds a usable open image, which routes around the institutional gaps.

The gap is real; the mitigations are honest. When an article references a Beijing-held object that bodhi cannot show, a related Met or Cleveland object is shown and the comparison is named.

Site photography and freedom of panorama

For sculpture in situ at Longmen, Yungang, Mogao exteriors, Nara temples (Tōdai-ji Daibutsu, Kōfuku-ji Asura), Kamakura Daibutsu, and Mt. Kōya:

Japan — Article 46 of the Japanese Copyright Act permits reproduction of artistic works permanently installed in open public places, including commercial use. Underlying Buddha sculptures are pre-modern and out of copyright; the photographer’s copyright is what governs the photograph. bodhi uses Wikimedia photographs licensed CC0 or CC-BY 4.0 by preference.
China — Article 24 (revised 2020) permits reproduction of works permanently installed outdoors. Longmen, Yungang, exterior Mogao, and Maijishan exteriors qualify. Permits and fees are point-of-capture concerns, not downstream-reproduction concerns affecting Wikimedia files already taken.
Korea — Article 35 allows reproduction of permanently-installed public works. Sculpture at Bulguksa is widely photographed; Seokguram’s interior is glass-enclosed and photography is restricted; the Haeinsa Tripitaka Koreana woodblocks are extremely restricted. The National Museum of Korea’s KOGL Type 1 holdings cover most of what bodhi needs to cite outside in-situ work.

The practical workflow is to search Commons by site category, prefer CC0 and CC-BY 4.0 over CC-BY-SA (which contaminates derivative-work licensing), enforce a minimum 2,400-px long edge, and cross-link to Wikidata for the site and the specific cave or sculpture.

Scholarly reading pool

Every iconographic article cites named scholars with page references. The core readings, organised by cluster:

Kannon and Mikkyō. Cynthea Bogel, With a Single Glance: Buddhist Icon and Early Mikkyō Vision (University of Hawai’i Press, 2009) is the load-bearing English-language reading for the Esoteric Kannon programme and the Six Kannon doctrinal placement. E. Dale Saunders, Mudrā: A Study of Symbolic Gestures in Japanese Buddhist Sculpture (Princeton / Bollingen, 1960) is the standard reference for per-form mudra readings. Hisashi Mōri, Sculpture of the Kamakura Period (Heibonsha / Weatherhill, 1974) anchors the Sanjūsangen-dō Senju Kannon and the Tankei chief image. Roger Goepper’s Aizen-Myōō: The Esoteric King of Lust (Artibus Asiae Supplementum 39, 1993) is the principal Aizen monograph; Bernard Faure’s The Fluid Pantheon (Hawai’i, 2015) treats the Fudō / Acala programme.

Pure Land and raigōzu. Mark Blum, The Origins and Development of Pure Land Buddhism in Japan (Oxford, 2002) for the doctrinal arc through Genshin, Hōnen, and Shinran. Robert Sharf, Coming to Terms with Chinese Buddhism (Hawai’i, 2002) for the Chinese-precedent reading. James Foard’s work on Hōnen and the Soga, alongside Esben Andreasen’s Popular Buddhism in Japan (Hawai’i, 1998), for the institutional reception of rapid-descent imagery.

Kamakura sculpture. John M. Rosenfield, Portraits of Chōgen: The Transformation of Buddhist Art in Early Medieval Japan (Brill, 2011) for the rebuilding programmes at Tōdai-ji and Kōfuku-ji. Samuel C. Morse on the Kei school. Mōri (as above) remains the indispensable period-wide reading.

Cross-cutting — Asuka, Hakuhō, transmission, technique. Donald McCallum, Hakuhō Sculpture (Hawai’i, 2009) for the seventh-century foundation. Sherry Fowler, Murō-ji: Rearranging Art and History at a Japanese Buddhist Temple (Hawai’i, 2005) for the temple-as-institution reading. The Metropolitan Museum’s Heilbrunn Timeline of Art History essays are used as cross-institutional curatorial anchors, not as a single-source authority.

Adrian Snodgrass, The Matrix and Diamond World Mandalas in Shingon Buddhism (Aditya Prakashan, 1988) is the structural reading for the Five Wisdom King / Five Wisdom Buddha pairing.

The reading pool deepens with each new article; this list names the load-bearing references rather than enumerating every cited author. Source-disagreement notes appear in the article body when the scholarship is contested.

Primary texts

Where iconography descends from a textual canon, bodhi cites the canonical text directly with edition and translator named.

Lotus Sūtra — Leon Hurvitz, Scripture of the Lotus Blossom of the Fine Dharma (Columbia University Press, 1976; rev. 2009) is the cited English translation. Taishō 262.
Mahāvairocana Sūtra (Dainichi-kyō) — Stephen Hodge, The Mahā-Vairocana-Abhisaṃbodhi Tantra (RoutledgeCurzon, 2003) for English; the Chinese-text edition Taishō 848 is cited by Taishō number for the Shingon-tradition reading.
Sarvatathāgata-tattva-saṃgraha (Kongōchō-gyō) — Taishō 865, Amoghavajra translation; principal source for Gōzanze Myō-ō and the Maheśvara-subjugation episode.
Pure Land Sūtras — Larger Sukhāvatī-vyūha, Smaller Sukhāvatī-vyūha, Kanmuryōjukyō (Contemplation Sūtra) — Hisao Inagaki, The Three Pure Land Sutras (BDK America, 1995/2003) is the cited English edition.
Ōjōyōshū (往生要集) — Genshin, 985 CE — the doctrinal anchor for the Nijūgo Bosatsu programme. James C. Dobbins’s reading is the cited Western scholarly entry-point.

Taishō Shinshū Daizōkyō numbers are given for sūtras whose digital open editions exist. Two databases carry these editions:

CBETA — Chinese Buddhist Electronic Text Association (cbeta.org).
SAT Daizōkyō — University of Tokyo (21dzk.l.u-tokyo.ac.jp).

Both are linked as research resources; neither is reproduced.

Source-language fidelity

Japanese terms carry Hepburn macrons (Jūichimen, not Juichimen); Sanskrit terms carry IAST diacritics (Avalokiteśvara, not Avalokiteshvara); kanji and kana are given on first mention, alongside the romanization. This is editorial, not decorative — the reader who wants to verify a claim against a Japanese-language catalogue must be able to see the source-language term in the body of the article.

Authority links

Every entity primer (figure, sculpture, painting, technique, concept, tradition) carries a Wikidata Q-ID as a stable external authority link. Where reciprocity is editorially appropriate, bodhi files a P973 (“described at URL”) statement on Wikidata pointing back to the entity page. The aim is a verifiable cross-link in both directions — Wikidata for authority, bodhi for editorial reading — without bodhi duplicating Wikidata’s structured-data role.

What bodhi does not cite as authority

Wikipedia article prose is not cited as authority. Wikipedia is used to discover scholarly references and to cross-check obvious facts; the citation chain runs through the named scholar or the primary text behind a Wikipedia claim, not through the Wikipedia article itself.
Popular-market trade books (introductory paperbacks without named-scholar peer review) are not cited.
Auction-house catalogue blurbs are cited for provenance only — the chain of ownership of a specific object — never as iconographic or art-historical authority.
AI-generated summaries and encyclopedic aggregations are never the source. The named scholar, the named primary text, and the named museum accession are the sources; research tooling may aggregate and surface them, but the citation chain runs through the underlying authority every time.

Updates and corrections

This page is reviewed each time the licence terms of a primary collection change or a new collection enters the reading pool. The full per-source protocol — when to cite, how to anchor a claim against which source, when to flag an attribution disagreement — lives in the project’s editorial standards, public for inspection in the agent-os/ directory of the project.

Corrections to attribution or licence claims: see the Image rights page for the takedown and correction procedure. Editorial corrections (source disagreements, attribution arguments, primary-text translation choices): file a GitHub issue with the article URL, the contested claim, and the source basis.