Institutions and tools
One corner of one city turns out to be held in a surprising number of hands, and it is only fair to say whose. Two archives recur throughout the documents used here. From the national library's digitised newspaper archive came a century of daily and trade press. The city's archive, Stadsarchief Rotterdam, supplied reading-room records, its image bank and address books, and with the household-register cards behind them these give the building its residents, its permits and its earliest photographs. Behind those two comes a longer tail of institutions, each with a particular kind of record: the national archive, the national heritage agency's photographic collection and the country's central family-history centre among them.
Rights are judged item by item, never inferred from an institution's name. An item for which the project has recorded no rights statement stays marked "not stated; local research only" and is not treated as cleared. A document that carries its own terms keeps them visible. Freely licensed images are the clearest case: some carry a Creative Commons licence, from the open media library, the European heritage portal and the city archive's own openly licensed pictures. When such a licence names a maker, the credit is given exactly as it requires. When the licence is a public-domain dedication instead, no credit is owed and none is claimed.
Open platforms made the record findable or reusable. A genealogical index opened the civil registers by name, and a web archive kept copies of the venue's own site after it had changed. A discography catalogue documented what was released from the building. An open media library supplied freely licensed images, a European cultural-heritage portal gave access to others, and a national art-history institute placed the painter in his profession. None of them holds the corner's story on its own, but each opened a part of it that would otherwise have stayed closed.
Automated language and vision systems helped with many first readings: they transcribed printed pages, described photographs and searched for documents by matching both their words and their meaning. Handwriting takes a stricter path. It is located automatically, then read blind in an independent double reading, each reader shown the crop cold with no candidate answer offered. The designated arbiter then weighs the readings against the scan, so no single machine is trusted. and explain the steps. These systems are acknowledged here for what they are: tools that did a first pass. Every reading they produced is a finding aid to be checked against the source, never a source in itself. The resulting text remains the project's interpretation. Naming a tool does not transfer authorship or prove that its output was accurate.