Digital resources in the Social Sciences and Humanities OpenEdition Our platforms OpenEdition Books OpenEdition Journals Hypotheses Calenda Libraries OpenEdition Freemium Follow us

Forgeries and Networks (ForNet): data transformations

Betende Frau nach Dirk Bouts. Links Vorbild, rechts Nachbildung.

by Jaap Geraerts, Henry Keazor, Demival Vasques Filho, Rebecca Welkens and Thorsten Wübbena

“Members are obliged to keep the Association’s printed matter carefully and secretly. They also undertake to return all printed matter to the Executive Board if they leave the Association and to ensure that the printed matter is returned to the Executive Board in the event of their death.”1

This clear call for secrecy and discretion can be found as a kind of prologue in issues of the so-called ‘Mittheilungen des Museen-Verbandes’, which were published from 1899 to 1939.2 They were distributed by the “International Association of Museum Officials in Defence Against Counterfeiting and Improper Trade Practices” (“Internationaler Verband von Museumsbeamten zur Abwehr von Fälschungen und unlauterem Geschäftsgebaren”). The project “Forgeries and Networks (ForNet) – The Mittheilungen des Museen-Verbandes and forgery networks in the 20th century”, funded by the German Research Foundation (DFG), will use this valuable source to reconstruct, visualise and analyse the networks of relationships and the strategies of the actors involved at the time as well as the routes taken by the forgeries and ensure that the data compiled in the project is made available to the public. This blog post, highlights the creation of the dataset, its transformations and ingest in a graph database, and the creation of networks, as a step towards answering the project’s research questions.

The “Verband” was founded in 1898 at the instigation of the Hamburg art historian Justus Brinckmann and the Zurich art historian, textile merchant, collector, and first director of the Swiss National Museum in Zurich, Heinrich Angst. Its members were mainly recruited from holders of higher positions within museums (directors or curators working as heads of collections), and for the time highly international; in addition to Germany and Switzerland, the places of origin of the two founders, Austria-Hungary and its successor states, Denmark, France, Great Britain, Italy, Netherlands, Norway, Spain, Sweden and the USA were also represented (the total number of members increased from 25 at the beginning to 411 in 1936.).

At the association’s annual conferences disputed objects were discussed, attributions questioned and technical investigations ordered. Some works were later rehabilitated by the exhibiting museums, while others were banished to the depots or remained in the display collections, where they are now once again erroneously listed as originals (more on this in a blog post at a later stage of the project). In the individual issues of the ‘Mittheilungen’, its members informed each other about current forgery cases and the state of knowledge about older and more active forgery workshops. In this context, clear warnings were also issued:

‘Our members have been aware of the forgeries and frauds of Max Heimann in Münster i. W. for several years’,3

‘I have been informed by an absolutely reliable source that Dr Erich Burg-Berge, who recently claimed to have discovered a Rubens in Antwerp according to newspaper reports, […] is a swindler and impostor’4

‘In any case, Mengaroni is one of the most dangerous copyists of all kinds of old majolica’.5

From a media-historical point of view, it is also worth mentioning that the works/forgeries mentioned were not only given an archive number to make them easier to recognise and identify, but were often also illustrated and frequently compared with correspondingly informative images of the originals.  In this way, the association has shaped art-historical knowledge to this day, without explicitly appearing in public (remember the call for discretion mentioned above).

In the project “Forgeries and Networks (ForNet) – The Mittheilungen des Museen-Verbandes and forgery networks in the 20th century”, funded by the German Research Foundation (DFG), one of its aims is to reconstruct a chapter in the history of art historical knowledge that has received little attention to date and to place it in the context of current debates on the role of museum institutions in the production of knowledge. This blog post, however, will not give an overview of the project as a whole, but rather zooms in on a particular aspect: the creation of a dataset and the several transformations of the data so as to make it amenable to being ingested in a graph database and the creation of networks and, in turn, to help answering the aforementioned and related research questions.

Creating a dataset

Although the ForNet project focuses on a variety of primary sources, the journal launched by the Verband – the Mittheilungen des Museen-Verbandes – takes the central stage. At present, a digitized copy of the Mittheilungen is hosted at the servers of Heidelberg University Library and the full ORC-generated text can be searched. While such a “string search” is useful, it does not allow one to subject the data to more detailed queries (e.g. forgeries of a particular type of object held in Austrian museums). For this (semi-)structured data is necessary. In order to create this, the six student assistants working for the ForNet project produce XML transcriptions containing the relevant information in each of the 812 entries of the Mittheilungen according to a custom-designed XML schema. Customarily, the artworks, people, and institutions, among other entities, mentioned in the entries are captured as well as the links between them.

While the creation of these XML transcriptions is a herculean task – with some transcriptions numbering hundreds and, in rare case, even thousands of lines – and are a valuable outcome in themselves, in the end they “merely” serve as a stepping stone towards the next steps undertaken in the ForNet project: the creation of networks based on the data in the XML files as well as the ingestion of this data into a graph database.

Graph database

While the data in the XML transcriptions is semi-structured, searching the XML files themselves is not incredibly helpful. Hence it was decided to ingest all the data into a graph database which allows one to subject the nodes (the main entities in our data) and the edges that connect them (the links in our data) to detailed queries. In order to do so, however, several steps need to be taken. To start with, via a Python script the data will be exported from XML to CSV files. Secondly, these CSV files need to be cleaned up (e.g. the duplicates have to be removed). Thirdly, not all our data is stored in the XML transcriptions but in several Excel tables as well. For example, the unique IDs of the different entities are stored in Excel files. These files need to be checked and enhanced on the basis of additional research (e.g. adding information about the nationality, religious affiliation, and sex of the members of the Verband). Fourthly and lastly, when all this work is done, via a series of import scripts the data will be imported from these files into the graph database where it is “reassembled” and “reconnected” (i.e. the links between the different entities in our data are restored).

Because the data is saved in different places this is a rather exacting and fiddly process, but in order to be able to examine the data systematically it needs to be stored and made accessible in one place, and the graph database does exactly that. The image below gives an impression of what the data in the graph database might look like. “Might” because this is very much work in progress and certain things such as the names of the edges can and probably will change along the way. The image nevertheless shows the level of detail that is captured and the wealth of data that our graph database will contain, thus allowing the creation and execution of detailed and granular queries.

The section of the graph database shows a network between entities of authors, owners, appraiser, recipients and several events and pieces.
Section of the graph database

Networks

As mentioned, the XML transcriptions also stand at the basis of a variety of networks that will be produced by the ForNet project. Created by means of intricate Python scripts, these (projected) networks zoom in on particular aspects of the data by reducing its dimensionality. For instance, a network might comprise particular entities in our data and the links between them. Whereas the graph database includes all the data contained in the XML transcriptions as well as the data stored elsewhere, the networks intend to open a window on particular subsets of our data. To offer one example, the network visualization below consists of the artworks, institutions, and people in our dataset and allows one to spot particular clusters that might be worth examining into more detail. Depending on our data and our particular research questions, various networks will be created, each of them offering a unique perspective on our dataset.

A large network graph of artworks, institutions, and people
Network graph of artworks, institutions, and people

Combining analogue and digital approaches

While the XML transcriptions, the graph database, and the networks are important and valuable outcomes in themselves, they do not stand on their own nor do they exist in splendid isolation. For the patterns that can be detected by using digital tools and methods – patterns that, due to the size of the data, can be tricky to be observed and are easily overlooked in an analogue environment – aim to foster and shape the archival research that is conducted by the ForNet project as well. By showing which people or institutions were intimately connected to the collection, production, and sale of forgeries, for example, the digital tools and methods provide focus and a sense of direction to the archival research. Simultaneously, this research will enrich and/or complement our dataset, thus putting flesh on the bones of our digital analyses, as it were. In the end, the various data transformations outlined in this blog aid the constant interaction and conversation between ForNet’s digital and analogue components, the very core of this project.


Featured Image: Praying woman after Dirk Bouts. Archiv 492. 493 Mitteil. 317.
https://doi.org/10.11588/diglit.56957#0017


Cite this article as: Jaap Geraerts, Henry Keazor, Demival Vasques Filho, Rebecca Welkens and Thorsten Wübbena: Forgeries and Networks (ForNet): data transformations. DH Lab (Blog). 26.06.2025, https://dhlab.hypotheses.org/?p=7237

 

  1. The original text reads as follows: “[…] sind die Mitglieder verpflichtet, die Drucksachen des Verbandes sorgfältig und geheim aufzubewahren. Sie verpflichten sich ferner, im Falle ihres Ausscheidens aus dem Verband die sämtlichen Drucksachen an den Vorort zurückzugeben, sowie dafür Sorge zu tragen, daß im Falle ihres Todes die Drucksachen an den Vorort zurückgegeben werden.”, example: https://doi.org/10.11588/diglit.35249#0070 []
  2. Available as a digital copy at Heidelberg University Library, https://doi.org/10.11588/diglit.33011, based on the collection held in the Kunstbibliothek, Staatliche Museen zu Berlin, https://www.smb.museum/en/museums-institutions/kunstbibliothek/home/ []
  3. https://doi.org/10.11588/diglit.35223#0032 []
  4. https://doi.org/10.11588/diglit.35248#0063 []
  5. https://doi.org/10.11588/diglit.35245#0023 []

OpenEdition schlägt Ihnen vor, diesen Beitrag wie folgt zu zitieren:
DH Lab (27. Juni 2025). Forgeries and Networks (ForNet): data transformations. DH Lab. Abgerufen am 23. Januar 2026 von https://doi.org/10.58079/1483o


Schreibe einen Kommentar

Deine E-Mail-Adresse wird nicht veröffentlicht. Erforderliche Felder sind mit * markiert

This site uses Akismet to reduce spam. Learn how your comment data is processed.