Provenance research: making an object’s past verifiable

Separate authorship from an object’s ownership history

The U.S. National Gallery of Art describes provenance as a work’s ownership history from creation to the present and continues to investigate owners, transfer dates and transaction types. Research also examines display, custody and use. A painting’s exhibition at a museum establishes a display relationship; it may have been a loan, so the museum cannot simply be added to the ownership chain.

Fictional situation: a ceramic label says “formerly in a painter’s collection,” while material analysis suggests a nineteenth-century date. Those claims concern collecting and manufacture; neither alone identifies the maker. Separate attribution, findspot, custody and legal title before asking which evidence answers each question. An attractive provenance statement cannot carry every conclusion.

A document still has to refer to the same object

Receipts, estate inventories, dealers’ ledgers, exhibition photographs and auction catalogues preserve different kinds of information. Record who made a document, why it was made, where it is held and exactly what it says. Secondary scholarship can guide the search, but five articles citing one catalogue do not provide five independent confirmations.

Fictional situation: a ledger lists “Riverside Landscape,” the title used today, but that alone does not establish a match. Compare dimensions, medium, reverse labels, old images, repairs and identifiers, allowing for changed titles, recording errors or measurements that included a frame. Setting agreements beside contradictions advances the inquiry more effectively than selecting one convenient line.

A timeline should show events and preserve gaps

A useful timeline distinguishes an appearance in a photograph, a claimed purchase, a documented sale and an inheritance. If a transfer is known only to have occurred between 1920 and 1930, retain that interval instead of inventing a year. A gap is missing information, neither proof of wrongdoing nor proof of lawful circulation. A negative search also needs a description of its scope.

The Van Gogh Museum’s 2017 annual report records that two paintings were stolen in 2002, recovered in Italy in September 2016, and formally returned and displayed again in 2017. Theft, recovery and return are different event types. They cannot be flattened into ordinary transfers between new owners, nor do they establish every movement during the missing years.

Digital searching expands reach and creates new blind spots

Getty launched a redesigned Provenance Index in 2025, using linked open data to connect historical records such as inventories, auction catalogues and dealers’ stock books. Such tools make it easier to trace people, objects and events across collections, but every database has geographical, chronological and documentary boundaries. It is not a complete register of every artwork worldwide.

Fictional situation: “Müller” may appear as “Mueller,” while optical character recognition may mistake the numeral 1 for the letter I. Try name variants, original-language terms and translations, then inspect the scanned page. Record the search date, query and database scope so that “not found” becomes an interpretable result, rather than a certificate that historical risk is absent.

Linked records and maps must not turn inference into fact

Networks of people, places and objects can reveal recurring galleries, shipping routes or collecting circles. Every connection still needs a relationship type: ownership, sale on behalf of another person, exhibition loan, or merely mention in a letter. Identical names may belong to different people, while one person may have several spellings. Stable identifiers and source references support checking rather than replace judgment.

Fictional situation: two databases import the same unverified collection history into a map, making the screen appear to show corroboration. Trace the shared source, label inferred connections and preserve the original wording. A timestamp, file hash or blockchain record can help document when data existed or whether it changed; none independently verifies the historical truth of the original entry.

Material research and provenance narratives should challenge each other

Pigments, panels, canvas, working marks and imaging can test claims about dates or techniques. Compatibility with a period does not establish that a particular artist made the work, and it does not determine whether later transfers were legitimate. A report should separate observations, analytical methods, comparative material and attribution inferences so another specialist can identify what can be checked again.

Christie’s 2017 timeline for Salvator Mundi is a useful exercise in reading qualifiers: parts of the early commission and ownership account are expressed as possibilities, alongside discussion of restoration and infrared examination. This is the auction house’s provenance and attribution account. Its hypotheses should not become an established royal ownership chain, and technical observations should not be presented as the final answer to attribution debate.

Bolton Council in the UK confirms that the Amarna Princess it acquired was a forgery by Shaun Greenhalgh. Institutional acceptance therefore does not make a conclusion immune to revision. The productive questions are which records supported the earlier assessment and what later evidence changed it. Retaining previous judgments and reasons for revision turns an error into a record from which others can learn.

Put a transaction back into its historical power relations

The manual prepared by the German Lost Art Foundation and partner institutions addresses Nazi persecution between 1933 and 1945: a recorded sale could still occur under discrimination, forced departure or dispossession. Research must examine transaction conditions and the parties’ circumstances, rather than stop at a receipt. The manual’s historical scope should not be applied unchanged to every colonial collection or other conflict.

For other objects, local names, oral histories, ceremonial uses or community memory may supply context absent from institutional archives. Collaboration can involve source communities, families, archivists, language specialists and conservators. Accounts need not be forced into a single narrative. Record who contributed them, how they were checked and where disagreement remains, while respecting conditions on sensitive knowledge.

A research report is not transaction clearance

Object provenance, attribution, legal title, border requirements and source of purchase funds intersect but are not interchangeable. FATF’s art and antiquities report discusses financial vulnerabilities involving intermediaries and concealed identities. A long, credible collection history does not itself establish who controls today’s buying company; an identified payer does not fill a gap in an object’s past.

A provenance narrative may affect market confidence, but it carries no fixed price premium. If the commissioning party intends to sell, disclose the research scope, payment relationship, unavailable archives and unresolved questions. Restitution and other rights disputes require separate assessment of relevant jurisdictions, dates and applicable rules. Distinguish what the research establishes, what can be legally claimed and what ethical action may be appropriate.

Leave a record another researcher can follow

A usable report identifies the object, provides a referenced event table, records archives searched, and states unresolved questions and confidence levels. A public summary can invite leads; access to fuller attachments can reflect privacy, contracts, copyright and cultural sensitivity. Preserve versions and reasons for revision instead of replacing the old account with a smoother story that conceals what changed.

Research can provide shared information for dialogue. WIPO distinguishes mediation, in which a neutral helps parties seek agreement, from arbitration, in which an arbitrator decides the dispute. Suitability depends on agreement, authority, the issue and applicable law; neither guarantees a quick settlement. Historical acknowledgment, return, access to records and continuing collaboration can be distinct concerns. Clarifying them helps evidence support concrete decisions.