Zurich Archives Discover Thousands of Duplicate Images Distorting Digital Collections

Zurich's two largest publicly accessible image repositories took coordinated steps this week to address a years-old problem that archivists have long flagged but rarely had the tools to fix: duplicate photographs catalogued under different reference numbers, sometimes with contradictory metadata, quietly polluting search results and misleading researchers who treat digital records as authoritative.

The Stadtarchiv Zürich, based at Neumarkt 4 in the Niederdorf district, confirmed it is running a full deduplication audit across its digitised photographic holdings this month. ETH-Bibliothek on the Rämistrasse campus is conducting a parallel review of its e-pics image portal, which serves researchers across Switzerland and internationally. The two institutions are coordinating their methodology, according to documentation circulated at a June workshop on Swiss cultural memory infrastructure.

Why This Week's Work Matters Beyond Housekeeping

Duplicate image replacement is not simply a tidiness exercise. When a photograph appears twice in a public database — once under a 1920s construction date and once attributed to the 1950s — every downstream use of that image inherits the confusion. Journalists, urban planners and historians at institutions from the Universität Zürich to the cantonal planning office on Stampfenbachstrasse draw on these collections when tracing the development of neighbourhoods like Aussersihl or the Kreis 5 industrial corridor. A single incorrectly dated photograph of, say, a Langstrasse building facade can ripple through academic papers and planning documents for decades.

The problem has grown with digitisation speed. Swiss archives accelerated scanning programmes substantially after 2020, when pandemic restrictions pushed researchers online. Volume increased faster than quality control could keep pace. By the end of 2024, the e-pics portal at ETH-Bibliothek held more than 2.7 million images, a figure the library itself cited in its 2024 annual report. The Stadtarchiv's own published inventory listed roughly 580,000 digitised photographic items as of early 2025. Even a duplication rate of two percent across those combined holdings produces tens of thousands of problematic records.

The specific trigger this week was a flagged batch of approximately 340 photographs of Zurich's Hardbrücke and the surrounding Industrie-Quartier, where multiple scanning rounds — conducted in 2018, 2021, and again in 2024 — had produced overlapping records. Some images appeared under three separate catalogue identifiers. Curators identified the cluster during a routine cross-reference in late June. The correction work began on Monday, 30 June, and archivists expect to clear the Hardbrücke batch by 11 July.

What Happens When a Duplicate Is Found

Replacing a duplicate is not as simple as deleting one copy. Swiss archival standards, governed by guidance from the Schweizerische Archivdirektorenkonferenz, require that any record touching a public collection be retired with a redirect rather than simply removed. This preserves citation trails for anyone who has already referenced the old identifier in a published work or legal document. The chosen canonical version is then updated with consolidated metadata, and the retired record carries a note pointing to it.

For the institutions involved, the workload is real. A single image with conflicting metadata can require correspondence between multiple departments to establish the correct attribution, particularly for photographs donated by private families. The Stadtarchiv has a dedicated processing team of six archivists; the ETH-Bibliothek digital collections unit is larger but also serves a much wider portfolio of scientific and technical material beyond photographs alone.

Researchers who rely on either portal are advised to re-check any citations to Hardbrücke or Industrie-Quartier photographs dated between 2018 and 2024, as reference numbers in that batch may change before the audit closes on 11 July. Both institutions plan to publish updated finding aids on their respective websites once the current round of corrections is complete. A joint technical note describing the deduplication methodology is expected to be released through the Swiss digital heritage network, Memoriav, by the end of the month — which could serve as a practical template for smaller cantonal archives that have accumulated similar backlogs without the staffing to address them.