The desideratum as well as diverse attempts to establish an overview of the dispersed royal treasure of Benin has existed for a long time, and has been present in academic circles (Dark, 1982) and activist demands since the 1970s and 1980s. After Nigeria’s independence, calls for the return, loan and/or restitution of the objects looted in Benin were often accompanied by requests for museums to supply listings of their holdings (see Bodenstein, 2022; Savoy, 2021). Recent developments in online museum catalogues has made it possible to access information on individual institutional holdings in certain cases. However, this access remains difficult for several reasons: first, the objects were part of general catalogues, requiring prior knowledge of object locations, and additionally, such catalogues are in different languages, many different search queries could be required to get a full list and museums often do not publish object information online. For the first time since their violent translocation, the Digital Benin platform shows the objects in one unique space accompanied by information that in certain cases had only partially been available to the public (i.e., internal museum data) or available in a dispersed form. With funding generously provided by the Ernst von Siemens Kunststiftung, the MARKK opened project offices in Hamburg and Benin City and in remote work environments in October 2020 for the two-year project time frame. The Digital Benin team sourced information, photographs and documentation material from institutional collections worldwide and extended the knowledge with oral histories and Digital Benin-authored research. In November 2022, the project launched online, providing access to a sustainable online archive. It will make a significant contribution to communicating and disseminating knowledge about an outstanding chapter of African art and cultural history to a broad and diverse audience. The platform establishes a unique resource, which seems all the more relevant at a time when many museums are preparing to restitute artefacts to their places of origin. Digital Benin will be transferred to a main host in Nigeria and will play a role in fostering further research, especially for Nigerian scholars, who are presently disadvantaged by the difficulty in accessing research materials and sources held in European and American museums and archives.
The project outline was completed after a two-day workshop organised at the MARKK in 2019, which brought together the necessary international expertise to prepare a comprehensive plan taking into consideration the needs of a wide range of users and audiences for a digital platform. The platform would be designed to make all identified works accessible, forming a digital hub for the myriad of object information, images, research materials and publications, many of which were not yet available online. It became clear that Digital Benin would also have to expand beyond the access to institutions’ information and digital material to also contextualise these with oral histories collected in Benin City and surrounding areas, the place of origin of the objects presented on the platform. Digital Benin aims to bring an Edo-centric focus to the embeddedness of knowledges, traditions and histories of the objects. This online knowledge forum creates a unique resource by connecting object data to historical contexts prior to the objects’ worldwide dispersal following the infamous ‘Benin punitive expedition’ in February 1897. This is ensured by presenting perspectives from practitioners and local specialists in Benin City and throughout Nigeria. The online platform presents different entry points into the study of the objects: Eyo Otọ, a foundational learning space about Benin objects and their Edo designations; a catalogue for searching and filtering institutional data for over five thousand objects from 131 institutions and twenty countries; oral histories of Edo people contextualising objects, sharing traditions and reflecting on Benin history; a map to explore historical and present-day sites of Benin Kingdom and the current location of the translocated objects in institutions worldwide; a list of the 131 institutions currently holding Benin objects in their collections; provenance for the study of roles, biographies and object relations of provenance names found in the information provided by the institutions and Itan Edo (story of Benin Kingdom) highlighting the socio-economic trajectories of various historical figures. A bibliography shows publications, reports, sales catalogues and more which are cited across the platform, and the media space includes 3D objects, educational video materials and documents to print.
The project began with a hiring process to fill the roles necessary for a complex digital project consisting of researchers, developers, designers, data stewards and a project catalyst (Langmead et al., 2018). The Digital Benin team is led by the principal investigators, Prof. Kokunre Agbontaen-Eghafona, Dr. Felicity Bodenstein, Prof. Dr. Barbara Plankensteiner, Dr. Jonathan Fine and Dr. Anne Luther, who is also the project catalyst leading the development of the platform on the intersection of research and technology. Design development is led by the studio Behavior/time, comprised of Alex Horak and Krystelle Denis; data processing and preparation is executed by Gwenlyn Tiedemann; Imogen Coulsen focuses on object research; Eiloghosa Obobaifo leads research in Edo; and Godfrey Ekhator leads oral history research in Benin City and surrounding areas. Ermeline de la Croix works on the data preparation and research on provenance names, and Mabel Oviahon supports content creation for the Benin Kingdom map. Research and development are woven into every step of the project and inform each other. The team works online in Nigeria, Germany, the United States and France, and meets regularly for workshops, team meetings and work sessions. Working through the pandemic on such an international scale meant bringing together distinct workflows and communication platforms to form an agile work plan. Research ideas and development are constantly negotiated, and decisions are made on accomplishable milestones in the time frame of the project. The project is growing with the data the team receives from the museums, and research is built inductively – the rich documentation on objects coming from museums is informing research and development decisions, as is the collection of multiple perspectives in videos and recordings. All team members areƒ sharing their expertise with one another, and effective communication between technical and research expertise has developed throughout the two years. The translation of many aspects of applied methods (such as technical feasibility, documentation or referencing) among the diverse team is made possible through a conscious decision to spend time on fostering good communication and a common language.
Digital Benin received information on over five thousand objects from 131 museums in twenty countries and digitised approximately 1,200 catalogue cards from the national museums in Benin City, Lagos and Owo where object information is stored in an analogue database. The transferred data ranges from images and 3D scans to information about the objects’ provenance, condition and curatorial research. The process of transferring and receiving data from the museums was based on building relationships with curators and directors of the contributing museums and is a testimony to existing forms of international collaboration, such as the Benin Dialogue Group, to bring together the data corpus on the Benin holdings. The list of museums grew from initial references identified in publications such as Benin: Kings and Rituals by Barbara Plankensteiner (2007), The Brutish Museums: The Benin Bronzes, Colonial Violence and Cultural Restitution by Dan Hicks (2021), An Introduction to Benin Art and Technology by Phillip Dark (1973) as well as referrals by the principal investigators, curators and museum researchers participating in the project. We set up remote meetings with curators and digital-collection staff to introduce the project, our data needs and negotiate licensing agreements.
The main technical consideration for the platform was to prioritise Nigerian users – hence we developed the platform showing diverse digital heritage material in a mobile-phone-friendly way. Design and interaction decisions, database development, server location and file formats were elements that needed to be recalculated and conceptualised for mobile internet access. According to a recent report, ‘In 2021, the number of unique mobile internet users stood at 4.32 billion, indicating that over 90 percent of the global internet population use a mobile device to go online’ (Ceci, 2022), and ‘In Nigeria if you look at users aged 16 to 64, 99.5% own a mobile phone with 99.2% owning a smartphone’ (Gilbert, 2022). Digital Benin aims to include communities that explore online spaces predominantly through mobile technologies and we built the site as light as possible without compromising the user experience for an inclusive exploration as part of decolonizing practices in digital heritage.
After the first six months of the project, we had acquired an initial data corpus that allowed us to develop strategies for a metadata mapping process to connect the datasets. The premise of the project is to preserve the data structure of each institution by displaying their distinct data setting of information on the platform while giving our users the possibility to search across all datasets. This data-mapping approach also resulted in a metadata structure that can be referenced in future digital projects in the museum context. So far, thirty-six comparable metadata fields transpired within 132 data sets, each with different data structures, languages and data standards. Partnering museums transferred between four and 191 data fields from their internal databases. Two-thirds of the transferred data comes from the ten museums that hold the most objects in their collections. Over 12,500 images were transferred by the museums with at most five hundred images for one object. Not all images match references in the datasets because some institutions store images in different database systems. A software was developed by the Digital Benin technical lead that generates a report for each institution of images that could be matched as well as missing and unmatched images. The image-matching process was an elaborate and time-consuming part of the data-preparation process for the platform.
The development team built twelve internal software tools that allow the researchers to build connections between datasets and objects based on in-depth research into the variety of objects and identification as well as exploration of museum standards for describing objects in databases and the development of controlled vocabularies in Edo and English. The controlled vocabulary shows, on the one hand, the original spelling and terms of each museum without cleaning or standardising the data, and, on the other hand, the development makes the search results more precise across all datasets by grouping distinct objects, provenance names and other terms. Thus, our users will be able to interact with an optimised search allows them to find objects in the multitude of spellings and languages that are found in the data and connect to all linked object information. The approach of mapping metadata across museum data schemas and creating a controlled vocabulary that groups similar terms answers two long-standing questions in digital-heritage development, namely how to connect museum data without changing internal data standards and how to make (re)search possible across different languages, terms and spellings. This approach is scalable and will be developed further by members of the team.
Digital Benin shows that it is possible to present multiplicity in digital cultural heritage and connect institutional data silos without imposing data cleaning or authoritative data standards. The data presented is published with each museum’s copyright and licensing, which shows that Digital Benin is a precursor to ‘digital restitution’, a field currently being defined by members of the team externally to the project. The steps that lead to digital restitution are taking inventory (Research Data Alliance International Indigenous Data Sovereignty Interest Group, 2020), negotiating ownership and ethical digital distribution (Odumosu, 2020) and building digital and legal infrastructures (Lixinski, 2020). Making data accessible is not yet digital restitution (Pavis & Wallace, 2020), and therefore Digital Benin is not a digital-restitution project but rather a precursor to possible negotiations of a full transfer of ownership over the digital material presented on the platform. The Benin objects show that the physical object lives in the context of songs, oral traditions, historical narrations – a plaque, for example, can make a viewer hear or sing a particular song, remember an Edo proverb and narration or pay tribute to a ceremonial site. Objects do not exist outside of this cultural tradition and presence; they are part of it. Digital Benin portrays the interconnectedness of multiple ways of viewing and understanding an object. The accumulated and interconnected digital heritage is integral to the life of the physical objects.
The project was organised in three distinct phases: research, prototyping and implementation.
Setting Up the Project
In the first weeks, the starting point was to acquire hardware for the offices in Hamburg and Nigeria, set up computers, email systems, sharing software such as Slack, Airtable, and Google Drive and a local server. Acquiring internet access in Nigeria as well as equipment for oral-history recordings required travel to Lagos from Benin City, which sometimes meant we had to work around limited handware availability in local stores. Additionally, we established workflows for the data collection from institutions, internal communications, consent and trust for the oral-history interviews and digitisation in the Nigerian museums. These processes were tied both to project management and to building trust internally as well as with our institutional partners and contributors. Below are some exemplary parts of the data-acquisition process:
Data Acquisition, Benin, Lagos, Owo: Introductions of the project and team members, visits to sites, digitisation of catalogue cards, triaging data, data transcription.
Data Acquisition, Museums: Introductions of the project and team members, sending out invitations, letters of intent and further informational material, negotiating legal agreements and licensing, receiving preliminary data that institutions can transfer immediately, receiving core and additional data.
Data Acquisition, Oral History: Outreach and research, introduction, acquiring hardware, first round of interviews, continuing outreach and interviews.
Phase 1 Research:
The two main goals – to build a data feed of differing data structures and an interface that allows elaborate research – were achieved in this phase. The data feed was built in a way that makes the interaction of the interface with the server as ‘light’ as possible, meaning that we worked with queries that have not been directed towards data on the server but rather that have been optimised for web interaction.
Metadata Mapping and Setting Our Data-Transfer Approach
We mapped around 1,500 field titles across 153 datasets and produced a standard for the most common fields in museum databases. Some datasets that we received are more complex than others, and we had to flatten relational data for the inclusion into the online catalogue. The metadata linking process was conducted inductively, meaning we studied the data sent to us to develop connecting categories based on similarities of the values in the respective fields. Similar to the historical process of digitisation practices in the 1970s (Sher, 1978, p.169), the examination led to thirty-seven fields that showed similarity across the datasets. All metadata fields that are not linked are still accessible in the full text searches across the datasets because they are connected by at least one field, namely the object IDs. The linking process was manually conducted, because the values in some fields did not allow an automatic linking process. For example, fields such as the credit line could show copyright information in one institution and prior owners in another. The manual examination of over 120 datasets from international institutions could be a starting point for a semi-automated mapping system that would allow the connection of diverse data in an online database for museums.
Approaches of standardisation for metadata and vocabularies to create a ‘single information system which embraces all museum holdings’ (Waibel et al., 2010) was first introduced in 1969, and as Lincoln pointed out, ‘There are several relevant authorities out there … VIAF, LOC subject headings, DBpedia, GeoNames, not to mention the Getty’s AAT, ULAN, and TGN. The British Museum links to none of them. […] They already have their own internal thesaurus of people and materials and subjects, and to map those to existing authorities would be a monumental task’ (Lincoln, 2016). Examining the data structures and vocabularies of datasets that were exported from internal databases as well as publicly accessed allowed us to review standards, vocabularies and cataloguing practices, and to come to the conclusion that no two datasets consisted of the same vocabularies or data structures. Conversations with curators, collection managers and IT staff in the beginning of the project gave the data team an insight into the data-production processes in the institutions. We received reports and indexes from the institutions and recommendations for using certain standards that were either institutionally used or a current trend in certain regions. The conversations, reviews of schema indexes and examination of the datasets showed that the approach of standardising all datasets would remove valuable insights about the local knowledge production of each museum. This also led to the decision to show fields that are not populated. We want to represent the full data context in which each institution produces knowledge about objects in their collections.
The approach in this project is based on Loukissas’ (2019) study on the locality of data in institutions in a sense that shows a certain placement within the logic of the collection (Groĭs, 2021) and hence the dataset/database it was retrieved from. The premise for building the platform was to ‘meet the data where it is’, which meant rather than to develop a new database with unifying data standards our project supports the data diversity produced by specialists in differing domains and maintains local data environments and non-standardised data – important elements for contextualising the data’s content and understanding the narratives it tells. We took up the challenge of producing an interface that connects non-standardised museum data by developing (or via the development of) a connection layer that makes it possible to connect diverse data provided by the institutions. Rather than asking our institutional partners to use the same metadata descriptors or database structures, we asked them to transfer the data as it is stored locally. In 2021, the German Contact Point for Collections from Colonial Contexts  asked Digital Benin for contacts and a list of institutions in Germany to collect data about Benin holdings in the country. The approach here was to quickly put together a database with simple and standardised data by asking the institutions to fill out a spreadsheet with seventeen fields (which some institutions did not fill out completely). This means that certain information available in the Digital Benin platform is missing from this database.
Another important factor for choosing the local approach to acquire data from the institutions was the proposition to hand over the data to a Nigerian partner at some point in the future. We organised the storing of data received from the museums in a way that would make a digital restitution process possible in the future, which would conclude the transfer of ownership of digital materials currently stored by the institutions about the Benin holdings. Our internal database has all data transferred by institutions stored in its original formats and documented with metadata about the transfer and data. Digital Benin is not a substitute for the restitution of the physical objects and is not changing the licensing or ownership of digital materials on the platform. This meant that we were asking for high-resolution images and digital material (including historical photographs of the objects). This process has stimulated a discourse about digital restitution and knowledge rights. For each institution, Digital Benin shows when, how and under what license data was received. Many institutions added a link to their own online-collections website, and therefore a linked reference to updates in their own online databases is included in the platform. In the beginning of the project, we proposed a semi-automated data pipeline which would allow updates made through API accesses, however we had to pivot away from this approach early on because it was not feasible to establish this pipeline for less than ten museums. However, the metadata mapping is a precursor for such a semi-automated online database, which could allow for a direct data feed into an online platform in the future.
After the considerations and data study outlined above, Digital Benin deducted thirty-seven field titles for the connection layer for the meta-data we received:
Inscriptions & Labels
Prior Source Role
Reign Associated Name
Department and Institution
The data preparation for the project was developed on the metadata and data levels that we received from the institutions. The data preparation was done as follows: data transferred in the way institutions store the data in their internal and published databases; preparation of data to be included in Airtable and immediately accessed by the researchers in the team; matching images with data from the institutions (image IDs did not automatically match for some institutions); and data extraction for prototyping in our internal research software.
Object Research and Itan Edo:
Phase 2 Prototyping
It took one and a half years to obtain the majority of data for the project, and after the initial six months, while we were still receiving data, we started to develop prototypes for research and the interactions on the platform. After assembling an exemplary data corpus, we set the scope for interactions and their feasibility. We defined the research areas as provenance, oral history, digitisation of data from Nigerian museums – which grew into a foundational space for the introduction and contextualisation of object groups in Ẹyo Otọ and historical/socio-cultural space on the map – and Itan Edo on the platform. During online workshops in the first phase of the project, the team developed the concept of thematic ‘doors’ for the prototype, which would allow users to open and walk through a series of doors to explore the research assembled in the different associated spaces. The idea was to create spaces with main doors, such as the door for the provenance space, and doors would connect these spaces to other spaces. Our users would be able to access the list of institutions through the catalogue or the object groups through the provenance space.
The question of how to make research material authored by Digital Benin accessible became another starting point for prototyping. It is obvious that the objects do not exist in a vacuum or as floating objects in vitrines. For example, they are related to the institutional provenance, including how they came to the museum, as well as their cultural and socio-political stories and indigenous contexts. This led us to the concept of knowledge weaving to show the multiplicity of values and meanings inherent to the objects. We not only focussed on an art-historical approach but embedded the knowledge production on the platform in a format that would allow us to render multiple ‘entry points’ to view, explore, understand, admire and interact with the digital material associated with the objects.
The doors and weaving concepts became stable communication tools within the team and opened the discursive workflow between the technical team and the researchers. Our technical team developed internal software tools so the researchers could immediately interact with the data, which allowed the team to develop workflows between multiple platforms (mostly Airtable, the internal software tools and prototypes). The learning curve for the team was steep, and the time investment for setting up clear communication, self-conducted workflows and shared digital workspaces was unique to this project, especially considering that some of our team members had not worked in fully digital spaces prior to this project. The expertise and skills in digital workflows, the use of multiple platforms and agile teamwork were by-products of the development of this platform.
In December 2021, we came together for the first time during the pandemic for a hybrid workshop in Hamburg at the MARKK. We organised visas, COVID testing, hotels and our conference room with the tireless support of the MARKK right before the opening of the exhibition Benin: Looted History. During the workshop, the platform’s doors were refined and links between each research space were clearly defined. The weaving process between the spaces started, creating an elaborate relational knowledge space. Two types of links were conceptualised: links intentionally curated by the researchers and hyperlinked text elements set by the developers semi-automatically. The prototyping and research process for each space is described in more detail here:
Eyo Otọ [Foundation]: Digitisation in NCMM Benin, Owo, Lagos
Phase 3 Implementation
In May 2022, we organised a workshop in Benin City to begin the last phase of the project, which meant the prototypes had to be finalised for the conceptual content production and desired interactions. The prototypes became iterations that went through constant review between the needs and anticipated interactions of the users that were set out in the first and second phases and the Digital Benin team’s collaboration between the research and technical teams.
We hired and consulted experts who carried out specific implementations. Amin Motallebzadeh prepared the sound quality and file formats for the media, Chao Tayiana prepared a document of recommendations and proposals for the display of derogatory, racist and harmful language on the Digital Benin platform and Osaze Amadasun became a design consultant and visual creator for the platform at large including the illustrations in the Eyo Otọ and map icons. The copy-editing of the texts was conducted by Max Bach.
Based on Chao Tayiana’s proposals, pop-up trigger warnings and suggestions in the catalogue were developed. Osaze Amadasun’s map icons were specifically designed for the content presented in the context of Benin. He also designed printouts: illustration cards that can be used as access points into each object group through QR cards and a colouring book with outlined illustrations of each object group.
The last phase concluded with a new iteration of the catalogue with filters, optimised free text searches, loading speed, an expanded and contained view of search results and links to the research spaces, among other things. Colour schemes, fonts and general design decisions were among the last things implemented, as the finalised content and research would inform these decisions. We consulted with Osaze Amadasun for the overall design of the platform, and he created multiple iterations for the site and icons specific to the interactions on Digital Benin. His design decks advised on placements of user experience and interaction choices, colour schemes, icons, lines and more to bring together the research and development. The last phase showed the potential to extend the project into a long-term research and field of study.
Extension of the Project
In July 2022, we received three hundred thousand euros from the Ernst von Siemens Kunststiftung to continue our work for another year. During the project, fifty institutions sent us historical or archival material, and while sorting it and beginning our research we realised that the material needs more detailed consideration and extended treatment. The nature of the material, with images of the violence caused by the looting in 1897 or letters written shortly after, as well as personal information and documentation of offensive and harmful language, requires a different approach to publishing the material. We will work on the accessibility and digitisation of the archival material until the fall of 2023. This project has great potential to become a long-term research endeavour, and many approaches it outlines are a baseline for further research and development.