(1) Overview
Repository location: https://doi.org/10.18743/DATA.00000405
Context
The Mapping Museums Lab is an interdisciplinary research team based at Birkbeck, University of London.1 Over the past decade, we have collected and analysed data to better understand the history, geography, and composition of the UK museum sector (Candlin et al., 2022). Our most recent project, “Museum Closure in the UK 2000-2025”, was funded by the Arts and Humanities Research Council and ran from 2023 to 2025.
The project investigated which museums had closed since 2000, why they had closed, and what had happened to their collections. Not all closures are alike. Some museum closures can result in a reduction in public amenities. In other cases, a local authority may amalgamate several museums into one larger and more efficient organisation, or replace a museum with a new visitor attraction and in such instances, services may be improved. Some museums sell their collections, some put them into deep storage where they are largely inaccessible, others transfer collections to another organisation where they are well used. Collections may be kept together or split between many recipients at the point of closure. Our aim was to clarify whether closures indicated a change in museum provision or a loss, and which collections stayed in public circulation, and which disappeared. A detailed analysis of this data was published as part of a report (Candlin et al., 2025).
(2) Method
Steps
During an earlier research project, the Mapping Museums Lab compiled a database of all the museums open in the UK between 1960–2020. This included information on the museums’ accreditation status, governance, location, size, subject matter, when they opened and if relevant, when they closed.2 We used the database to establish an initial list of museums that closed between 2000–2025 and their attributes. We then collected extensive new data on more recent and additional closures, the reasons why they had closed, and on the subsequent disposal and dispersal of collections. Wherever possible, we gathered information on who acquired objects from the closed museums, under what circumstances, and where the objects were moved.
Our data came from a variety of sources, including: local and national newspapers; the Museums Journal; online searches; staff at museums, museum services, the Museums Development Network, and local authorities; subject specialist networks and journals; Charity Commission and Companies House records, including liquidators’ reports; closed museums’ websites (extant and via the Wayback Machine3); social media; online reviews (primarily TripAdvisor4), blogs, and chat forums; enthusiast, preservation, and historical societies; neighbouring businesses and tourist offices.
The data for collection disposal and dispersal, and reasons for closure were entered into a spreadsheet used as input to a Neo4j graph database. The full details of the data modelling process, the database, and its use in our research are provided in Wright et al. (2025).
To increase interoperability, each museum was assigned its unique ID from the Mapping Museums database. The disposal and dispersal of collections were recorded as events, each with sender and recipient actors and involving either an object or a collection. Our data model shares some overlap with CIDOC Conceptual Reference Model (CRM), an ontology for cultural and heritage objects (Bekiari et al., 2024).
Collections or parts of collections involved in post-closure events were assigned unique IDs; to facilitate reuse, the types of objects contained in them were categorised using Wikidata items; when relevant, named individual items with an entry on Wikidata were also linked to their ID.
Events entailing a change in ownership or custody were assigned an actor recipient. Each actor was assigned a unique ID; a type (e.g. museum, local authority); a sector (e.g. public, private); and an address. Actors that are museums in the Mapping Museums database were also assigned their Mapping Museums ID.
Quality control
In the initial data collection, we followed the terminology about event and actor types used in the original data sources. Subsequently, we normalised and standardised the terms into a controlled vocabulary of event and actor types and then structured these into reusable taxonomies (see Appendix 1, Tables 2 and 3).
As dating closures and the related disposal and dispersal events is often difficult, uncertain dates of events were encoded using the Library of Congress’ Extended Date/Time Format (EDTF).5 Similarly, where the type of an event was uncertain, this was indicated by appending a question mark to the event type.
During the data collection process, data sources sometimes contained conflicting information or lacked details about certain museums or collections. If this were the case, members of the project team would try to contact relevant staff from the closed museum in order to clarify what had happened. Often, multiple members of the project team would contact multiple sources so that information could be cross-checked and validated. However, this information cannot be shared as part of the dataset as it contains personal details.
Once data had been collected and inserted into a spreadsheet, we developed a bespoke software tool to convert the data into a Neo4j6 database according to a graph data model developed by the research team.7 The upload tool examined the data to ensure it was written in the correct format and used the correct vocabulary. For example, it checked that dates were compliant with EDTF and that actors and events had types defined in our taxonomies.
(3) Dataset Description
Repository name
BiRD – Birkbeck Research Data
Object name
dispersal_events.csv, event_types.csv, actor_types.csv, closure_reasons.csv
Format names and versions
CSV
Creation dates
2023-10-02 to 2025-11-01
Dataset creators
Andrea Ballatore (King’s College London, United Kingdom): data modelling
Helena Bonett (University of Exeter, United Kingdom): data collection
Fiona Candlin (Birkbeck, University of London, United Kingdom): data conceptualization, data collection, and data modelling
Maria Golovteeva (Warburg Institute, London, United Kingdom): data collection
Mark Liebenrood (Birkbeck, University of London, United Kingdom): data collection and data modelling
Alexandra Poulovassilis (Birkbeck, University of London, United Kingdom): data modelling
Peter T. Wood (Birkbeck, University of London, United Kingdom): data modelling
George A. Wright (Birkbeck, University of London, United Kingdom): data modelling and data curation
Language
English
License
Creative Commons Attribution–NonCommercial 4.0 International License (CC BY-NC 4.0).
Publication date
2026-01-08
(4) Reuse Potential
Beyond its immediate role in documenting museum closures, this dataset can contribute to a broader understanding of the UK’s changing cultural sector and the shifting location and status of museum objects. The dataset offers significant reuse potential across disciplines: it can underpin longitudinal and spatial analyses of cultural infrastructure; support network analysis of important actors in the culture and heritage sector; and inform policy decisions about heritage funding and collections management.
A key limitation of the available data on museum closures is its variable granularity: some transactions are recorded at the level of single objects while others are recorded at the level of museum sub-collections or entire collections. This constrains the detailed tracking of individual objects over time.
Researchers can connect the data to other Linked Open Data resources: museums are referenced by their Mapping Museums identifiers which can in turn be found on Wikidata; the contents of collections are described using Wikidata items including reference to individual named objects where known; and 42% of non-museum actors are recorded with postcodes, enabling linkage to geographical and demographic datasets. Our bespoke taxonomies of event and actor types are not directly linked to Wikidata or other external ontologies and may require mapping onto external sources if these parts of the dataset are to be integrated with others.
The dataset also provides a valuable teaching resource for courses in data modelling and cultural heritage informatics.
Appendices
Appendix
Table 1
Fields included in the file dispersal_events.csv.
| COLUMN NAME | DESCRIPTION |
|---|---|
| initial_museum_id | The Mapping Museums ID of the museum where the object(s) involved in the event originate(s). |
| initial_museum_name | The name of the initial museum. |
| initial_museum_size | The Mapping Museums size category of the initial museum, estimated from its yearly visitor numbers – huge (more than 1 million visitors), large (50 thousand to 1 million visitors), medium (10 thousand to 50 thousand visitors), small (fewer than 10 thousand visitors), unknown. |
| initial_museum_governance | The governance category of the initial museum – national, local authority, other government, university, independent, private, unknown. |
| initial_museum_accreditation | The accreditation status of the initial museum – accredited or unaccredited. |
| initial_museum_subject | The subject matter category of the initial museum. |
| initial_museum_town | The town where the initial museum was located. |
| initial_museum_region | The country or English region where the initial museum was located. |
| initial_museum_postcode | The postcode of the initial museum. |
| super_event_id | Unique identifier assigned to each super-event (i.e. each museum closure). |
| super_event_date | The year of the museum closure. |
| super_event_reasons | A list of reasons why the museum closed (drawn from a controlled vocabulary structured into a bespoke type hierarchy – see Table 4). |
| event_stage_in_path | The stage in the sequence of events involving the same object(s) (1 if the event is the first recorded event, 2 if it is the second, etc.). |
| event_type | The type of event (e.g. sold-at-auction, sank). |
| event_core_type | A more general type that the event is an instance of (e.g. sold, damaged/destroyed). |
| event_is_end_of_existence | True if the event entailed the end of the object(s)’s existence. |
| event_is_change_of_ownership | True if the sender gave ownership of the object(s) to the recipient. |
| event_is_change_of_custody | True if the sender gave physical custody of the object(s) to the recipient. |
| event_date | The date of the event (in extended date/time format) if the event happened at a point in time. |
| event_date_from | The start date of the event (in extended date/time format) if the event took place over a period of time. |
| event_date_to | The end date of the event (in extended date/time format) if the event took place over a period of time. |
| previous_event_id | The ID of the immediately preceding event involving the same object(s). |
| sender_id | The unique identifier of the actor playing the sender role in the event. |
| sender_mm_id | The Mapping Museums ID of the sender if the sender is a known UK museum. |
| sender_name | The name of the sender. |
| sender_type | The type of the sender. |
| sender_core_type | A more general type that the sender is an instance of. |
| sender_sector | The economic sector that the sender belongs to – public, private, third, university, hybrid, unknown. |
| recipient_id | The unique identifier of the actor playing the recipient role in the event. |
| recipient_mm_id | The Mapping Museums ID of the recipient if the recipient is a known UK museum. |
| recipient_name | The name of the recipient. |
| recipient_quantity | The quantity of people or organisations this actor represents (a specific number, or many) |
| recipient_type | The type of the recipient. |
| recipient_core_type | A more general type that the recipient is an instance of. |
| recipient_sector | The economic sector that the recipient belongs to – public, private, third, university, hybrid, unknown. |
| recipient_size | If the recipient is a museum in the Mapping Museums database, its size category – huge, large, medium, small, unknown. |
| recipient_governance | If the recipient is a museum in the Mapping Museums database, its governance category – national, local authority, other government, university, independent, private, unknown. |
| recipient_accreditation | If the recipient is a museum in the Mapping Museums database, its accreditation status – accredited, unaccredited. |
| recipient_subject | If the recipient is a museum in the Mapping Museums database, its subject matter category. |
| recipient_town | The town where the recipient is located. |
| recipient_region | The country or English region where the recipient is located. |
| recipient_postcode | The postcode of the recipient. |
| object_id | A unique identifier assigned to the object(s) that the event involves. Unique identifiers are usually constructed by concatenating the super event ID and an abbreviation of the museum’s name with a short description of the object(s). Object IDs are not official identifiers. |
| parent_object_id | The ID of the object(s) that the object(s) was/were previously part of. |
| object_size | A description of the proportion of the museum’s original collection that the object(s) made up (all, most, half, some, few). |
| object_quantity | The precise number of object(s) when known. |
| object_status | The status of the object(s) – collection (objects from a museum’s collection), loan (objects on loan to the museum when it closed); handling (handling objects), museum-stuff (other items such as display cases). |
| object_types | A list of Wikidata items describing the types of object(s). |
| object_description | An English description of the object(s). |
Table 2
Fields included in the file event_types.csv.
| COLUMN NAME | DESCRIPTION |
|---|---|
| type_name | The name of the event type. |
| sub_type_of | The name of its parent type (if any). |
| is_core_category | True if this type should be treated as a core category (neither highly specific nor highly generic). |
| is_change_of_ownership | True if events of this type are assumed to entail a change of ownership of the object(s) involved. |
| is_change_of_custody | True if events of this type are assumed to entail a change of the physical custody of the object(s) involved. |
| is_end_of_existence | True if events of this type are assumed to entail the destruction of the object(s) involved. |
Notes
[1] https://mapping-museums.bbk.ac.uk (last accessed: 7 January 2026).
[2] See https://museweb.dcs.bbk.ac.uk/home (last accessed: 7 January 2026).
[3] https://web.archive.org (last accessed: 7 January 2026).
[4] https://www.tripadvisor.com (last accessed: 7 January 2026).
[5] https://loc.gov/standards/datetime (last accessed: 7 January 2026).
[6] https://neo4j.com (last accessed: 7 January 2026).
[7] The data model is detailed in full on the project’s GitHub respository: https://github.com/Birkbeck/museum-object-flows/tree/main/data-model (last accessed: 7 January 2026).
Acknowledgements
We gratefully thank the hundreds of people who provided information on their institutions and organisations, who gave advice or provided leads, or who checked our data. We also thank UKRI-AHRC for funding the project and the project’s external Advisory Board.
Competing Interests
The authors have no competing interests to declare.
Author Contributions
Mark Liebenrood: data curation, investigation, methodology, validation, writing – original draft.
George A. Wright: data curation, formal analysis, methodology, resources, software, visualization, writing – original draft.
Andrea Ballatore: conceptualization, funding acquisition, formal analysis, methodology, supervision, writing – review & editing.
Fiona Candlin: conceptualization, funding acquisition, investigation, methodology, project administration, supervision, writing – review & editing.
Alexandra Poulovassilis: conceptualization, funding acquisition, methodology, project administration, software, supervision, writing – review & editing.
Peter T. Wood: conceptualization, funding acquisition, methodology, project administration, supervision, writing – review & editing.