Abstract

Museums are not contextless institutions and their datasets reflect that. This project (1) examines how the Metropolitan Museum of Art’s information infrastructure obfuscates critical examination of its possession of Benin bronzes; (2) uses vis created using OpenRefine and Tableau Online tools to show how the asynchronism between continuous and discrete data fundamentally limits applying vis tools to object documentation in museums; and (3) offers direction for future research.

Introduction

The Benin Bronzes

Within museum dialogues, the phrase “Benin bronzes” refers to the estimated 3,000-5,000 plaques, ivory tusks, sculptures, and beautiful coral and brass and ivory objects that formed the royal collection at the Court of Benin. After a British so-called punitive expedition invaded Benin City in 1897 and plundered its palaces and shrines, many of these treasures were sold at unknown dates and under unclear circumstances into the burgeoning international market for ethnographic objects and works of art. The Metropolitan Museum of Art first acquired its current body of Benin bronzes from private collectors who donated their acquisitions to the museum. This is one the roundabout routes by which other Benin bronzes, taken illegally, have been displayed in the museums of the global North for generations. 

Calling the Met’s possession of Benin bronzes “controversial” would be a grievous understatement. International outcry has recently spurred institutions in the global North to consider alternatives to their continued curatorship of Benin bronzes. The proposed Edo Museum of West African Art in Benin City, Nigeria has been posited as a future home for the array of looted artifacts being returned to Nigeria by museums around the world, striving to reunite loaned “Benin artworks currently within international collections” while investigating the broader histories represented by these artifacts.¹ As of June 2021, the Met has confirmed that it will return three of its estimated 160 bronzes to Nigeria.²

“The Met is not saying these objects were stolen, so morally, legally we’ve decided to give them back. […] They want to be seen as willing and helpful and good partners to Nigeria.”

Barnaby Phillips, LOOT: Britain and the Benin Bronzes (2021)

Motivation

At first, I wanted to use Tableau to visualize the unclear quantity of how many Benin bronzes the Met currently retains in its collection, in order to convey how the Met remains largely unmoved to confront its complicity in colonialism. However, my thinking that I would be able to use the datasets available to the public to achieve this task was quickly derailed, primarily because the datasets didn’t exist in a usable form. As such, the original vis I set out to complete turned into an exploration of the ways that the Met’s public datasets obfuscate the repatriation of Benin bronzes.

I took inspiration from museum vis projects like Jer Thorp’s A Library of Color for Library of Congress, Colour Explorer and chronological histograms from Tate Gallery and the Centre for Australian Art (CAA), and DNBVIS (Deutsche Digitale Bibliothek Visualized) with the German National Library. Though these projects all utilize much more complex vis tools, I was very taken with how they all promoted the transparency and organized searchability of their collections. Color is a noticeable organizer in these largely visual collections; I appreciated the projects’ use of color to convey information and made sure to incorporate color to differentiate information in mine.

Tate and CAA’s use of interactive histograms rather than more complex moving elements to visualize large amounts of data was particularly helpful to me because I didn’t have access to tools with moving elements for this project. I wish I had been able to link the records themselves to their entries in my vis like CAA does, because I think that provides a very grounding point of reference for users who might otherwise feel lost in data. It’s also disappointing that there is so little data in the Met’s dataset about the creators of the Benin bronzes, because these examples of museum data vis do such interesting work with artist provenance.

Scope

I focused on the Met rather than the British Museum or Ethnology Museum in Berlin (or any other of the collecting institutions that house Benin bronzes) for an answer that is straightforward but feels unsatisfying in a pedagogical sense: the Met is one of the most prominent American museums with the largest repository of Benin bronzes still housed in its collections, with a public .csv dataset in English that I could clean up in the time I had available as a full-time grad student.³ I pursued this project primarily to practice creating vis using OpenRefine and Tableau on public datasets rather than to professionally build a deliverable product, so if I work professionally with similar kinds of projects in the future, I would love to be able to attribute them the time and breadth and depth of research they deserve.

All data used in this visualization is derived from the Met’s original dataset as of October 20, 2021.⁴ My project’s scope is limited by the fact that the Met’s datasets of information on more than 470,000 items are still ongoing and parts of the datasets are incomplete. The datasets are regularly updated with new and revised information on a regular basis: this means that as of my publishing this report, the data is as current as possible; however, it may become outdated with the future publication of a new dataset.

Methodology

Cleaning Data

After downloading the October 20, 2021 dataset from the Met’s Open AccessGitHub, I began to work with it in OpenRefine with the intention of a) gathering identifiers with variated names under one term, and b) reducing the amount of entity columns I had to transport into Tableau (for efficiency). OpenRefine made the first very easy: I removed columns that were irrelevant to my project or contained no data, creating a more concise dataset.

Challenges

When accomplishing the second task, I ran into several challenges due to the analog nature of the data. Continuous data is highly valued in museum object documentation because it will better communicate provenance information. This makes the way that museums document provenance inherently very messy, because in a qualitative setting like an object label in a display, this kind of variation improves accuracy. However, in a discrete activity like cleaning up a dataset, this variation hugely impeded accuracy.

Examples of this: the Met’s dataset used terms like “(?),” “or,” and “probably” to imply an unclear historical record rather than having a singular identifier for objects suspected to originate from the same place (i.e., Court of Benin). Additionally, the dataset didn’t ID specific artifacts past “Object Name” and “Title” entities (aka, no record if these objects were part of a contested collection or in the process of being returned). These two trends made it extremely difficult to coalesce variations firmly into distinct identifiers. In order to clean data up around this, I ended up grouping variations with the same intention so that I could usably work with them in Tableau. I wish I’d found a happy medium in preparing a discrete dataset for usability without reducing the importance of analog data in accurate provenance research, but I did not. Therefore, I remain dissatisfied with what I feel is a critical lack of nuance in the dataset I visualized.

Visualizing Data

Once I had my dataset usably clean in OpenRefine, I connected it to Tableau Online and began to see what stories I could pick apart. I ended up being able to essentially prove every detail I’d set out to find. The complicated part came at the end, when I tried to prove that records that fit every category of being the Met’s Benin bronzes were the Met’s Benin bronzes.

Key Findings

Vis: I combined the count of MetObjects.csv with “Accession Date” to track when the Met acquired Edo art.
Finding: The Met’s largest acquisition of objects with an “Edo peoples” identifier was in 1991.

Vis: I ran “Credit Line” against “Accession Year” to track from whom the Met accessioned the 1991 donation of Edo art.
Finding: The 1991 accession is credited to Mr. and Mrs. Klaus G. Perls.

Vis: I broke down the Perls’ donation of art and artifacts created by Edo peoples by object name to examine the contents of the accession.
Finding: The contents of the Perls donation were, by majority, plaques, bracelets, and sculpture heads.

Reaching these conclusions was like knowing what the puzzle looked like and how the pieces fit together, all at once, and not being able to call a puzzle solved. Benin bronzes largely consist of plaques, sculpted heads, and beautiful coral and bronze and ivory ornaments— this is not in doubt.

The Met has Benin bronzes on display and in its holdings— this is not in doubt.

Klaus Perls acknowledged that his collection contained Benin bronzes and donated it to the Met in 1991— this is not in doubt.

AND YET: The fact that I can conclude all the above using the Met’s datasets doesn’t mean I’ve proved that these objects were Benin bronzes. I’ve simply proved that the objects fit all the things we know about the Met’s collection of Benin bronzes. Without anything in the dataset to make the final link between these objects and Benin bronzes explicit, all I’ve done is infer.

Future Directions

The difficulty I had telling the story of the Benin bronzes using the Met’s open data reveals the fundamental limitation of applying vis tools to open datasets when it comes to object documentation in museums: in the words of La Tanya S. Autry and Mike Murawski, museums are not neutral. Museum datasets are not contextless and do not tell neutral stories. Museums, past and present, benefit greatly by claiming a sort of intellectual objectivity that they absolutely do not have. No matter how “willing” an institution comes off, they very rarely will accede on an artifact that cannot be quantitatively proven as rightfully not theirs. Institutions regularly cite lack of provable data as reasons to not return contended items. Datasets like the Met’s exemplify how easily information in open datasets can still be undocumented, even in conversations as global as these.

And the conversation truly is global. Digital Benin is leading an international effort to create a well-founded and sustainable catalogue of the Benin artworks and their history, cultural significance, and provenance. Sarr and Savoy’s groundbreaking 2018 report The Restitution of African Cultural Heritage. Toward a New Relational Ethics offers concrete steps towards restitution as a curative process for museums.⁵ The Benin Dialogue Group strives to establish a system of rotating loans of artworks to Benin City from a consortium of European museums.⁶ Academics, politicians, artists, community members, and museum stakeholders are pushing internationally for reparative actions from Western museums. I’m excited to see how data vis will be involved in this movement; hopefully, in the future, someone will be able to reconcile the continuous-discrete asynchronism of museum datasets better than I.

References

Bakare, L. (2021, March 26). Regional museums break ranks with UK government on return of Benin bronzes. The Guardian.

Ezra, K. (1992). Royal art of Benin: the Perls collection in the Metropolitan Museum of Art. Metropolitan Museum of Art.

Graze, M. (2020, September 1). Museums Are Going Digital—and Borrowing From Data Viz in the Process. Medium.

Greenberger, A. (2021, June 17). Germany Unveils Comprehensive Database of Its Benin Bronzes. ARTNews.

LaGamma, A. (2021, March 4). The Legacy of Benin Court Art: From Tragedy to Resilience. Metropolitan Museum of Art.

Lusher, A. (2018, June 24). British museums may loan Nigeria bronzes that were stolen from Nigeria by British imperialists. The Independent.

British Museum. (2020, November 13). Major new archaeology project on site of new museum in Benin. British Museum.

Prussian Cultural Heritage Foundation. Managing Non-European Objects. Prussian Cultural Heritage Foundation.

Phillips, B. (2021). LOOT: Britain and the Benin Bronzes. One World Publications.

Restitution Report 2018.

Seiff, A. (2014, July 1). How countries are successfully using the law to get looted cultural treasures back. ABA Journal.

Tchan, I. (2018, March 12). Restitution of Beninese heritage: Open letter to Felwine Sarr and Bénédicte Savoy. Mediapart.

Ethnologisches Museum. The Benin Collection in Berlin. Prussian Cultural Heritage Foundation.

Thorp, Jer. (2018). A Library of Colors. Medium.

Citations

¹ Gershon, L. (2020, November 17). A New Museum of West African Art Will Incorporate the Ruins of Benin City. Smithsonian Magazine.

² Greenberger, A. (2021, June 10). Metropolitan Museum of Art Returns Two Benin Bronzes, Signaling a Major Shift. ARTNews.

³ Brown, Kate. (2018, July 27). Benin’s Looted Bronzes Are All Over the Western World. Here Are 7 Museums That Hold Over 2,000 of the Famed Sculptures. ArtNet.

⁴ Metropolitan Museum of Art. (2021, October 20). Metropolitan Museum of Art Open Access CSV, GitHub. CC0.

⁵ Sarr, Felwine, Bénédicte, S. Trans. Drew S. Burk. (2018, November). The Restitution of African Cultural Heritage. Toward a New Relational Ethics.

⁶ Hohensee, Naraelle, Stuart, G. Benin Plaque: Equestrian Oba and Attendants. Khan Academy.