Method
A chain, not a label
The provenance of a work is the documented history of its ownership and custody: from the photographer’s studio to its present location, through every sale, gift, bequest, deposit, or seizure.
It isn’t a statement of origin. It’s a chain, and its value lies in continuity: each link names a counterparty, a date, a type of transfer and, ideally, a source attesting to it. A chain with a forty-year gap says something important — provided the gap is visible rather than papered over.
What it establishes, and what it doesn’t
It establishes the legitimacy of possession and informs authenticity: a work documented from the studio onward is hard to forge. It’s also what institutions require before a purchase or a deposit.
It does not establish authenticity on its own. A provenance can be accurate and attach to a misattributed work. That’s why provenance and attribution are two distinct registers, not to be merged into one free-text field.
How to record it usefully
Free text is easy to write and impossible to query. Structure each event:
- Type of transfer: sale, gift, bequest, deposit, inheritance, restitution.
- Counterparties, normalised as entities rather than strings — “Gallery X” must be the same object across all thirty records citing it.
- The date, at its real precision: a bare year is honest information, so is “circa 1962”.
- The source attesting to it: invoice, correspondence, exhibition catalogue, register.
The sensitive part: it’s almost never publishable as-is
A complete provenance contains purchase prices and the names of living buyers. Those are commercial and personal data. Publishing them by default in a report or an export exposes your collector and your market.
The habit that protects you is default deny: carried-over or imported provenance stays confidential until a human has decided, row by row, what is publishable. An empty section in a report can be fixed; a disclosed buyer cannot be taken back.
In practice
- Record events as structured entries, not as a paragraph.
- Normalise counterparties into reusable entities.
- Make gaps visible instead of filling them with assumptions.
- Treat confidentiality as default deny, released by review.
- Always attach the source that attests to the link.
Obscura Flow models provenance as dated events with normalised counterparties, and marks rows carried over from an older model as confidential until reviewed — publication is an act, not a side effect.