The museum is closed. We hope to see you soon. Sign up for our newsletter and we will keep you posted!

Longread — 17 Feb 2021

Introduction: In the Presence of Data

It is at first glance counterintuitive that In the Presence of Absence would include a study of institutional data in its exhibition’s accompanying text. The exhibition called for works “that can challenge existing or conventional ways of knowing” and knowledge forms historically excluded in official records, institutional frameworks, and other prevailing knowledge systems. Why, then, conduct a study of the museum’s own data and publish it in the context of this exhibition?

It is partly an issue of sustainability of the curatorial vision beyond the current exhibition. It is important to take inventory of the museum’s official recorded history to better understand what is present and absent in the archive. This was done by analyzing more than 100,000 object records and 35,000 artist records in Adlib, the collection management system.1 The data was processed and visualized using Microsoft Excel, custom Python scripts, and Tableau.

The curators wanted to research the completeness of information in the archive in relation to past acquisitions. While assembling their concept for In the Presence of Absence the curators noticed recurrent absences and errors in the museum’s collection data in Adlib. For instance, the original installation instructions for artist Hala Elkoussey’s work Mount of Forgetfulness could not be traced, a film about a man seeking to preserve disappearing stories—a troublesomely ironic example of archival omission. In other cases, records had incomplete fields or duplicate entries for individual artists. These absences and inaccuracies made it difficult to research specific works and artists, and thus to draw conclusive arguments about the collection. The curators were curious about what the information enclosed in the database could say about the collection, but also how information gaps could be identified and ultimately resolved. Notwithstanding the inaccuracies, the curators wanted to use data research to answer other critical questions about the collection.

Figure 1 (Summer 2020): Number of artist records attributed to different regions. The chart does not include region “Oceania (Excluding Australia and New Zealand),” which is composed of regions Melanesia, Micronesia, and Polynesia, from which there are no artist records. Regional groupings are based on the geographic regions defined under the Standard Country or Area Codes for Statistical Use (known as M49)5 of the United Nations Statistics Division. The numbers in this chart include duplicate records.

Research on Birthplace: Numbers and Context

Visualized in Figure 1, the vast majority, almost 91%, of artists who have their birthplace identified in their record (29% of total artist records) were indicated to have been born in either Europe or North America. Western Europe alone accounts for 64% of artist births, and 41% of artists overall were born in the Netherlands. The countries with the most artist birth records were the Netherlands (4,100+ records), Germany (846+ records), the United States (806+ records), and France (573+ records). Outside of European countries and North America, the country with the most birth records was Japan (240+ records) followed by Indonesia (146+ records) and then China (52+ records). Overall, Eastern and Southeastern Asia accounted for roughly 5% of artist births, Latin America and the Caribbean accounted for 2% of births, and Northern Africa and Western Asia accounted for about 1% of artist births. Many countries had no artist birth records, including Myanmar, the Philippines, Uganda, Jordan, Palestine, Oman, and Somalia. Countries including Syria, Peru, Iraq, Kenya, and the Dominican Republic had between 1-4 records for artists born there.

These counted values and percentages, however, can be misleading and are inconclusive in assessing the museum’s investment or reach in and beyond the Netherlands. As seen in Figure 6, 71% of artist records do not have complete birthplace information and are thus not included in these facts and figures. The visuals only capture part of the picture without the birthplace information of the majority of acquired artists. Furthermore, even if all birthplaces were reported, records containing geographic data in the form of birthplaces, places of death, and nationalities do not necessarily point to the places in which artists developed their practice or made/called their home(s). Counting artists based on their country of birth or listed nationality thus results in tentative and unclear metrics of representation and access, which do not address topics of race, ethnicity, migration, diaspora, redrawn borders, or self-identification. 

Regional birthplace maps are shown below in Figures 2-5. Each square represents a birth city of an artist and the size of the square is proportional to the number of birth records from that place. On the right side of the maps, there is a panel showing the nationalities identified in the artist records for artists born in the country in focus. As stated in the figure caption, the records counted in Figure 1 and in the maps include duplicate entries. Duplicate entries are when an artist has two or more records attributed to their name or a misspelling of their name. Based on about 10 country data sets of varying sizes, we estimated that duplicates make up less than 10% of artist records overall. In the case of artist birth records from Indonesia, as much as 17% of entries for artist records are duplicates. While the reason for this specific example is not clear, it appears that many duplicate entries often have partial information, which, if combined, would create a more complete single entry. Unfortunately, because of the number of records, it was not possible for a single data researcher to go through all of them, eliminate identical duplicates, or combine/synthesize the information in duplicates with differing/partial information, especially since this would require biographical archival research on individual persons.

  • Figure 2 (Summer 2020): Europe: Regional map for artist records.
    Figure 2 (Summer 2020): Europe: Regional map for artist records.
  • Figure 3 (Summer 2020): Latin America and the Caribbean: Regional map for artist records.
    Figure 3 (Summer 2020): Latin America and the Caribbean: Regional map for artist records.
  • Figure 4 (Summer 2020): Africa and Asia: Regional map for artist records.
    Figure 4 (Summer 2020): Africa and Asia: Regional map for artist records.
  • Figure 5 (Summer 2020): Asia and Oceania: Regional map for artist records.
    Figure 5 (Summer 2020): Asia and Oceania: Regional map for artist records.

Considering the Object Catalog

Looking only at biographical information in artist records resulted in a significant pitfall: not all artists have the same number of acquired art objects. For example, an artist with 271 works acquired like Kazimir Malevich would be counted in Figure 1 as being equally represented in the collection as an artist who has only a single work acquired. This is especially important to consider when analyzing gender disparities in the collection. As shown in Figure 7, there is a difference in assessing which percentage of artist records are identified as women versus how many artworks acquired in the collection were made by women artists. While 13% of artist records with identified birthplace were identified as female or vrouw, women are far less represented when the number of objects acquired is taken into account. Only about 8% of the works in the collection were made by a woman artist.

There were similar implications for the analysis of the artist's birthplace records when the number of objects acquired per artist was taken into account. For example, 91% of artists were born in Europe and North America, but 94% of objects in the collection were created by an artist born in that same region.

Figure 6 (Summer 2020): Analysis of birthplace information in museum database Adlib.
Figure 6 (Summer 2020): Analysis of birthplace information in museum database Adlib.
Figure 7 (Summer 2020): Gender Analysis of Artist Records with Birthplace Versus Object Records.
Figure 7 (Summer 2020): Gender Analysis of Artist Records with Birthplace Versus Object Records.

Note on Gender Documentation 

Roughly one-third (11,239 records) of artist records overall (34,609 records) have gender indicated. Important to mention is that within gendered records, gender is binary and inconsistently reported. In some cases, gender is reported in a field titled “Gender”, identified as “male” or “female” (48% of records). In other cases, gender is reported in a field titled “History/Biography”, identified as “vrouw*” or “man” (72% of records). There are cases in which gender is reported in both categories (overlapping in 20% of records). A regional comparison found that artists born in Latin America and the Caribbean are more likely to have gender identified in the “Gender” field (80% of records) than artists born in Africa whose gender is more often reported in “History and Biography” and only 23% of the time in the “Gender” field. Inconsistencies in reporting were initially confusing, and researchers should thus take both fields into account.

The Object Catalog: A Closer Look

To study the object catalog, a timeline of object records per acquisition year was created (Figure 8). The figure shows the development of the collection in the past 125 years of Stedelijk history. It reveals that 50% of the current object catalog was collected by 1988. The figure shows the portion of works created by women artists as well as the tenures of museum directors.

In Figure 9, each square represents an artist, and the area of the square is proportional to the number of works for that artist. Some remarks that need to be addressed for understanding the underlying data: This visual does not include the 8% of total object records made by unknown or “onbekend” artists. Neither does this study take into account the size or price of the works. For example, the study of the object catalog (93,042 works) led to several interesting findings, such as that Willem Sandberg (2,213 works) and Wim Crouwel (1,777 works) were the two most prominently represented persons in terms of the number of works. It is important to mention that although the representation of their work in the collection is numerous, this representation partly consists of paper invitations, posters, and other graphic expressions of the Stedelijk Museum, for which they both provided the graphic design. This leads to the urgency to emphasize that when discussing individual object records there is no taking into account the size or prominence of the works in question within this study. For instance, one artist could have many individual object records for preliminary sketches or small illustrations in books and newspapers whereas another artist could have a single object record for a large work. Furthermore, individual objects within an installation may be documented as separate records, if they are components of the same work but can also be considered as separate works. Overall, the work of the top 1% of most collected artists (artists with the most works) makes up a third of the collection. About 54% of artists have only one object collected, 83% of artists have 5 or fewer works collected. The most prominent women artists in the collection included graphic designer and typographer Fré Cohen (334 works) and textile artist Kitty van der Mijll Dekker (287 works). Data sets had to be cross-referenced, as the artist biographical data is separate from the internal object catalog. Errors such as misspellings or missing fields can impact the accuracy of the depicted data. For instance, in the timeline Figure 8, about two-thirds of the objects had an acquisition date included in the record, leaving one-third of objects unrepresented.

Many of the artists with the most works attributed to them were designers, which led to a curiosity about the composition of the collection by media which can be seen in Figure 11. Each art object has a sub-collection designation in the database based on its media (e.g., Photography, Works on Paper, etc.). This allows to create an overview of the museum collection bases on the type of art. Together, the top four largest sub-collections Typography (22%), Posters (20%), Photography (11%), Works on Paper (11%) contain over 60% of the works.

  • Figure 8: Number of acquired objects vs. year of acquisition in Stedelijk history. The number of acquired objects (purple and orange area) are represented by the left axis and the respective percentage is indicated on the right and traced by black line.
    Figure 8: Number of acquired objects vs. year of acquisition in Stedelijk history. The number of acquired objects (purple and orange area) are represented by the left axis and the respective percentage is indicated on the right and traced by black line.
  • Figure 9 (Summer 2020): Number of objects per artists represented tile plot. Each tile corresponds to one artist entry with the area being proportional to the number of object records attributed to them.
    Figure 9 (Summer 2020): Number of objects per artists represented tile plot. Each tile corresponds to one artist entry with the area being proportional to the number of object records attributed to them.
  • Figure 10 (Summer 2020): Artists and collectives with 50 or more object records attributed to them.
    Figure 10 (Summer 2020): Artists and collectives with 50 or more object records attributed to them.
  • Figure 11 (Summer 2020): Breakdown of object records by collection. Each tile corresponds to one collection with the area being proportional to the number of object records attributed to them.
    Figure 11 (Summer 2020): Breakdown of object records by collection. Each tile corresponds to one collection with the area being proportional to the number of object records attributed to them.

Conclusion & Outlook

In accordance with the curatorial objectives for In the Presence of Absence, this study examined the data available in the Stedelijk’s database. The study evaluated the data for completeness and identified missing information fields. In addition, the study used available data to answer questions about artist birthplace, artist gender, acquisitions over time, and the internal object catalog.

Based on the data processing and visualization, the study revealed these main results. Of the artists with birthplace indicated (29% of total artist records), the vast majority of artists were born in Europe and North America (91%), with 41% of artists having been born in the Netherlands. In studying the object catalog, the study found that half the current collection (based on number of objects) was acquired by 1988. It was additionally found that the largest sub-collections in the museum are Typography, Posters, Photography, and Works on Paper. This study did not evaluate monetary value of works, exposure of works in exhibitions, or research labor per artist or object. These are great topics for future research especially because taking financial aspects of acquired works into account could provide a better understanding of the museum’s investment in artists. Further research would profit from more consistent guidelines for data entry and record maintenance.

Investing in collection data research is an exciting means towards institutional self-awareness, understanding of historical context, and creating a legacy. Without archival conservation and interventions to correct errors and fill in missing information, many artists and works may be misrepresented or forgotten in the digital historical record. Included in the Stedelijk’s vision, as defined on the museum’s website, is the claim “Our museum builds memories for the future.” Revisiting the past in the archive, however, may be the boldest move forward.

Footnotes

[1] A Note on Adlib:

Adlib was a landmark achievement for the Stedelijk and the culmination of efforts towards collection record digitization that began in the 1980s. First, information on paper inventory cards was transferred to computer records. Consecutively, in 2000, the museum adapted Adlib as their official collection data system and since then, all new acquisitions have been directly put into Adlib. The object records vary in the depth of their documentation. Achieving 100% digitization of the collection, with 50% of the collection available for view online is a notable feat. There is, however, a considerable need for more data researchers and data entry professionals for the database’s upkeep to preserve history and correct errors.