# A Closing Date That Moves Both Ways, and Who Actually Moved It
> Of 937 exhibitions tracked here, only 7 ever had a date move, and three of those flipped between the same two dates again and again. The log says who moved them.
> The Study - 2026-09-24
> Canonical HTML: https://iconotheca.com/study/a-closing-date-that-moves-both-ways/
## Key points
- The exhibitions corpus holds 937 shows and 1,605 log entries, and only 54 of those entries record a date actually moving. They fall on seven exhibitions. Everything else is a status pass, a first listing or a correction.
- Three of those seven did not move once. They flipped between the same two dates repeatedly: the Acropolis Museum show 31 times between 30 August and 6 September 2026, sixteen one way and fifteen back.
- For two of the seven the archive recorded the cause in its own log, and the cause was the archive. A stale staged file was replayed four separate times by an unscoped merge, once because merge_exhibitions.py was called with --help, which is not one of its flags and so ran the full job.
- A date that oscillates is not a museum changing its mind. It is two readings of one page, and without a log that timestamps every read there is no way to tell that apart from a real postponement.
- The same ambiguity sits on a single object's page. The Mauritshuis gives one window for the painting being off view and a different, shorter one for the building being shut, and both are true.
## An exhibition date is a reading, not a fact
## 1,605 log entries, and only 54 of them are a date moving
## Seven exhibitions carry every date move in the corpus
## Thirty-one moves between exactly two dates
## Two shows at one museum exchanged each other's closing dates
## For two of the seven, the archive names itself as the cause
## The same ambiguity sits on a single object's page
## What a change log has to record to be worth keeping
## How to read an exhibition date
## Why this archive is free to read
Fifty-four date movements in this archive, and the seven exhibitions carrying them tell a story about the reader rather than about the museums. Three shows flipped between the same two dates again and again, and where the archive looked hard enough to find a cause the cause was its own tooling replaying a stale file. That is not a reason to distrust the calendar. It is the reason to keep a log that records the old value, the new value and the reason, because a museum that changes its mind does it once and says so, while a reading error does it over and over in both directions. Store the date you read and the day you read it, and the difference becomes visible.
## Questions
**How many exhibitions in this archive have had their dates changed?**
Seven, out of 937 tracked. Two of those changed once, which is what a genuine rescheduling looks like. The other five account for 52 of the 54 recorded movements between them.
**Why would an exhibition date move and then move back?**
Because the date is being read rather than announced. If one reading takes a date from a shared listing index and the next takes it from the show's own page, or if a stale staged file is replayed over live data, the recorded date will alternate between two values while the museum's actual plan never changes.
**Did this archive cause any of the date changes it recorded?**
Yes, and it says so in its own log. Four entries record an unscoped merge replaying a stale staged file over current data, affecting the Naples Parthenope show and the Harewood Cecil Beaton show. One entry names the trigger as a call to the merge script with --help, which is not a recognised flag and therefore ran the full job.
**Is a museum's closure the same as an object being off view?**
No. Read on 24 September 2026, the Mauritshuis listed Girl with a Pearl Earring as off display from 15 August to 4 October 2026 while the building itself was shut only from 24 August to 20 September. The painting is away for about two weeks after the museum reopens.
**What should I check before travelling to see a particular object?**
Check the object's own page rather than the museum's general opening hours, check the date on which that page was last confirmed, and prefer a ticketing page to an editorial one. An object can be off view while its museum is open, on loan while its catalogue entry still names its home room, or in a rotation that shows it for a few weeks a year.
## Sources
- [1] Mauritshuis, The Hague, collection page for Johannes Vermeer, Girl with a Pearl Earring, inv. 670. Read in the in-app browser on 24 September 2026 after a plain fetch returned HTTP 403 from the site's bot protection. Source for the two overlapping absences quoted here, for the dimensions and signature, and for the museum's published provenance line.
- [2] MACBA, Museu d'Art Contemporani de Barcelona, exhibitions index. The source URL recorded for Anna Moreno. The Third Twist, and a listing index carrying several shows at once.
- [3] Acropolis Museum, Athens, temporary exhibitions page for Inspirations. Ancient Greek Art living in Italy. The single source URL against which 31 date movements were logged between 24 August and early September 2026.
- [4] Harewood House, event page for Cecil Beaton: Staging Icons. The source re-confirmed directly after an unscoped merge reset the show to superseded dates.
- [5] The Walters Art Museum, object record 54.456, Ewer, 644 AH/AD 1246-1247, acquired by Henry Walters 1917. Read in the in-app browser on 24 September 2026 after HTTP 403 on the plain fetch. Source for the exhibition list quoted in the caption and for the museum's published provenance of three names.
- [6] The Metropolitan Museum of Art, Open Access Collection API, object 451753, accession 64.178.2. Queried directly on 24 September 2026. The response carries no objectHistory and no provenance field, and reports isPublicDomain true.