Gustav's Hand: Digitisation, Digital Enhancement, and Dissemination of the Gustavian Collection
By testamentary disposition, the private archive of King Gustav III (1746–1792) – comprising his correspondence and manuscripts on a number of subjects – was left in the care of Uppsala University Library (UUL) upon his death. This collection is essential for Swedish and international research concerning the eighteenth century, and researchers have used it for studies of the period’s political, social, and cultural life for centuries. This unique collection has put Uppsala on the map of eighteenth-century studies. Therefore, it is most unfortunate that UUL has deemed it necessary to block the originals from all use. Years of use have exacted a toll, and further use entails an immediate risk of the loss of invaluable texts. The project will rectify this situation and boost scholarship based on the collection, achieving this by four means: 1) the physical manuscripts will be conserved and digitised, and thus be made more available for sustainable use; 2) handwritten text recognition (HTR) will make principal parts of the collection searchable as text and thus vastly improve their accessibility; 3) research on the digitised material will be initiated, with the material’s searchability allowing new and large-scale research questions to be posed (e.g., through text mining), and research opportunities will be offered to junior scholars during the project; and 4) the collection will be disseminated in cultural heritage contexts through social media, programme activities, etc.
Final report
Through testamentary provisions, Gustav III’s private archive came to Uppsala University Library (UUL) after his death in 1792. The collection, today known as the Gustavian Collection, consists of approximately 200,000 pages of manuscripts—nearly twelve linear meters—divided into 125 volumes, of which 16 contain manuscripts written in Gustav III’s own hand. There are few equivalents to this cohesive archive from a royal individual, built up over nearly forty years—from the prince’s early school notebooks to the adult king’s speeches, letters, and various projects. Since its opening for consultation, the collection has been central to Swedish and international eighteenth-century research. However, years of use took a heavy toll on the material. By 2019, UUL was forced to restrict access to it. Continued use would have posed an acute risk of major damage and irreversible text loss. This situation was the starting point for the present project, whose goal was to create a sustainable research infrastructure for the collection by conserving, digitising, and making this unique archive accessible for future research.
PURPOSE
The project has had four aims:
1. to restore, conserve, and digitise the collection
2. to make documents written by Gustav III searchable through HTR
3. to initiate research and teaching based on the collection
4. to communicate the project and make the collection present within the cultural-heritage sphere
RESULTS
The final report follows the project’s four aims.
Conservation and Digitisation
The conservation of the collection has been completed with excellent results. The physical collection has undergone extensive conservation and storage measures to ensure its long-term preservation, and the material is now well prepared for future research and continued use. The access restriction on the physical collection has been lifted.
Personnel funds were reallocated within the project, enabling the recruitment of an additional conservator, which in turn facilitated the completion of the work in accordance with the established plan.
The entire collection has been digitised and published on Alvin, the platform for digital cultural heritage used by UUL and 20 other Swedish and Norwegian cultural heritage institutions. Particular focus was placed on Gustav’s autograph writings, which constitute a central part of the collection’s scholarly value. In total, approximately 1,200 documents in this part of the collection have been published as individual Alvin records. This fine-grained level of document description creates good conditions for future development of the records through expanded subject descriptions, supplementary registers, or integration of research results. The work, therefore, represents not only an effort to preserve the material but also to support future knowledge production. The remaining parts of the collection have been published at the volume level, in accordance with the project’s publication strategy.
The digital catalogue is, with respect to Gustav’s autograph writings, more detailed than the older paper catalogues. For instance, the place of dispatch for letters, where known, has been consistently recorded and linked to each locality’s authority record in Alvin. Authority-record links also exist for senders and receivers of letters. The digital catalogue further clarifies which documents are genuinely in the king’s own hand. There are, in fact, several copies and transcripts written by others. Information on which documents were added to the collection during the nineteenth century (the so-called Tersmeden donations, etc.) has also been consistently included wherever this is evident in older catalogues or in the documents themselves. A tag for the Gustavian Collection has been added to every Alvin record from the collection, allowing users to limit searches to that material.
The project’s primary aim—to secure the collection physically and digitally and to make it accessible in a sustainable and research-promoting manner—has thus been achieved. The foundation laid enables both deeper cataloguing and thematic research pathways, as well as future development projects involving the digital collection.
HTR
The work on Handwritten Text Recognition (HTR) has delivered strong results and, together with the digital catalogue, constitutes the most significant outcome of the project’s digital enrichment. Based on the base model “Stockholm Notaries 1700 2.1,” the project has developed a specialised HTR model, “Gustav III – SWE – older hand,” for Gustav’s Swedish handwriting. The model has undergone several rounds of training and has reached a Character Error Rate of 5.7%. The model is fully usable for efficient semi-automatic transcription and offers substantial time savings in the continuing work with the collection.
A functioning workflow, from material preparation to quality assurance and publication, has been developed, combining machine reading with manual correction via crowdsourcing and expert review. The first crowdsourcing event was held in spring 2023. Thereafter, events have been arranged each semester. A total of six events have been held with around fifty participants. Students have made up a large proportion of participants, but members of the public have also joined. These events have contributed valuable manual corrections while also helping to spread knowledge about the project and how HTR works in practice.
The Swedish documents in volumes F411, F415, and F418 have been used in crowdsourcing, and at present, 79 documents have been manually corrected, checked, and published with transcription in Alvin. Each such document includes a searchable PDF of the handwritten text, allowing users to search and select the transcribed text. The transcription is also available as separate Word and TXT files. The former allows those who cannot read the handwriting—or who want a quick overview—to access the transcription in a familiar format. The latter is useful for those who wish to analyze transcriptions in an open file format that enables large-scale analysis and interoperability.
In addition to the published transcriptions, a large number of documents have been started but require further manual work before publication. This work can be continued in follow-up projects, as both expertise and all working materials are available at UUL.
Research Initiation and Teaching
For administrative reasons, the project’s ambitions concerning student theses have only been realised on a limited scale; however, three master’s theses in history are currently in progress. It may also be noted that in 2023 the project was the subject of a master’s thesis in archival science at the Department of ALM, Lund University, and in 2025 a master’s thesis in the same field at Uppsala University.
The digitisation of the collection has undoubtedly—fully in line with the project’s intentions—made it far more accessible to both student and senior researchers, in Sweden and abroad.
The integration of the collection into teaching has been successful. The project has been presented to students in the international Master’s Programme in Digital Humanities (Department of ALM) as an example of cultural-heritage digitisation in spring 2023, 2024, 2025 (and continuing in 2026). For these students, a workshop on the project’s workflow was also held in autumn 2022 and autumn 2023.
The project has hosted a total of five student interns (three from History and two from ALM), who, within their advanced studies (7.5 or 15 credits), participated in the project and developed skills in machine reading and HTR.
Internationally, in spring terms 2022, 2023, 2024, and 2025, the project has contributed lectures and workshops to the course “Things That Matter: Materials & Culture in/for the Digital Age”, organized in collaboration between Uppsala University and the universities of Bern, Durham, Groningen, and Tübingen.
Outreach
The project has been communicated extensively through various channels. Updates and blog posts have been published on the website and on social media. A total of around 10 presentations have been given in relevant settings, including conferences, academic seminars, archives, museums, and other cultural heritage institutions. The project has also attracted media attention, appearing in newspapers, on television, on the radio, and in podcasts. An exhibition has been produced and shown twice in the Carolina Rediviva exhibition hall at UUL. In addition, the project has organised a lecture series with invited speakers—one per semester—at UUL.
The goal of making the collection and the project visible within the cultural-heritage sphere has thus been achieved.
The project has resulted in three publications:
• Mikael Alm & Johan Sjöberg, “Att bygga infrastruktur för forskning: Den Gustavianska samlingen vid Uppsala universitetsbibliotek,” in Henning Hansen & Peter Sjökvist (eds.), Från handskrift till digitalt kulturarv: Perspektiv på specialsamlingar, Uppsala, 2025, pp. 83–104
• Mikael Alm & Johan Sjöberg, “Dräkter för riddarspel och dramatik i Gustav III:s privata arkiv,” in Sofia Nestor (ed.), Teaterkungens kostymer, Stockholm, 2025, pp. 76–83
• Mikael Alm, “Det började med en skiss,” Historisk tidskrift (forthcoming 2026)
USE AND ACCESSIBILITY OF THE INFRASTRUCTURE
The project has created an infrastructure at UUL that can be applied to future projects. Establishing a workflow that spans the project’s components and units (conservation, image capture, HTR reading, accessibility, communication, and preservation) has been crucial for efficient work and avoiding bottlenecks. The outcome will benefit future projects at UUL.
The conservation and digital publication efforts are fully integrated into UUL’s operations. UUL is responsible for the collection and for Alvin. Alvin is publicly accessible.
PURPOSE
The project has had four aims:
1. to restore, conserve, and digitise the collection
2. to make documents written by Gustav III searchable through HTR
3. to initiate research and teaching based on the collection
4. to communicate the project and make the collection present within the cultural-heritage sphere
RESULTS
The final report follows the project’s four aims.
Conservation and Digitisation
The conservation of the collection has been completed with excellent results. The physical collection has undergone extensive conservation and storage measures to ensure its long-term preservation, and the material is now well prepared for future research and continued use. The access restriction on the physical collection has been lifted.
Personnel funds were reallocated within the project, enabling the recruitment of an additional conservator, which in turn facilitated the completion of the work in accordance with the established plan.
The entire collection has been digitised and published on Alvin, the platform for digital cultural heritage used by UUL and 20 other Swedish and Norwegian cultural heritage institutions. Particular focus was placed on Gustav’s autograph writings, which constitute a central part of the collection’s scholarly value. In total, approximately 1,200 documents in this part of the collection have been published as individual Alvin records. This fine-grained level of document description creates good conditions for future development of the records through expanded subject descriptions, supplementary registers, or integration of research results. The work, therefore, represents not only an effort to preserve the material but also to support future knowledge production. The remaining parts of the collection have been published at the volume level, in accordance with the project’s publication strategy.
The digital catalogue is, with respect to Gustav’s autograph writings, more detailed than the older paper catalogues. For instance, the place of dispatch for letters, where known, has been consistently recorded and linked to each locality’s authority record in Alvin. Authority-record links also exist for senders and receivers of letters. The digital catalogue further clarifies which documents are genuinely in the king’s own hand. There are, in fact, several copies and transcripts written by others. Information on which documents were added to the collection during the nineteenth century (the so-called Tersmeden donations, etc.) has also been consistently included wherever this is evident in older catalogues or in the documents themselves. A tag for the Gustavian Collection has been added to every Alvin record from the collection, allowing users to limit searches to that material.
The project’s primary aim—to secure the collection physically and digitally and to make it accessible in a sustainable and research-promoting manner—has thus been achieved. The foundation laid enables both deeper cataloguing and thematic research pathways, as well as future development projects involving the digital collection.
HTR
The work on Handwritten Text Recognition (HTR) has delivered strong results and, together with the digital catalogue, constitutes the most significant outcome of the project’s digital enrichment. Based on the base model “Stockholm Notaries 1700 2.1,” the project has developed a specialised HTR model, “Gustav III – SWE – older hand,” for Gustav’s Swedish handwriting. The model has undergone several rounds of training and has reached a Character Error Rate of 5.7%. The model is fully usable for efficient semi-automatic transcription and offers substantial time savings in the continuing work with the collection.
A functioning workflow, from material preparation to quality assurance and publication, has been developed, combining machine reading with manual correction via crowdsourcing and expert review. The first crowdsourcing event was held in spring 2023. Thereafter, events have been arranged each semester. A total of six events have been held with around fifty participants. Students have made up a large proportion of participants, but members of the public have also joined. These events have contributed valuable manual corrections while also helping to spread knowledge about the project and how HTR works in practice.
The Swedish documents in volumes F411, F415, and F418 have been used in crowdsourcing, and at present, 79 documents have been manually corrected, checked, and published with transcription in Alvin. Each such document includes a searchable PDF of the handwritten text, allowing users to search and select the transcribed text. The transcription is also available as separate Word and TXT files. The former allows those who cannot read the handwriting—or who want a quick overview—to access the transcription in a familiar format. The latter is useful for those who wish to analyze transcriptions in an open file format that enables large-scale analysis and interoperability.
In addition to the published transcriptions, a large number of documents have been started but require further manual work before publication. This work can be continued in follow-up projects, as both expertise and all working materials are available at UUL.
Research Initiation and Teaching
For administrative reasons, the project’s ambitions concerning student theses have only been realised on a limited scale; however, three master’s theses in history are currently in progress. It may also be noted that in 2023 the project was the subject of a master’s thesis in archival science at the Department of ALM, Lund University, and in 2025 a master’s thesis in the same field at Uppsala University.
The digitisation of the collection has undoubtedly—fully in line with the project’s intentions—made it far more accessible to both student and senior researchers, in Sweden and abroad.
The integration of the collection into teaching has been successful. The project has been presented to students in the international Master’s Programme in Digital Humanities (Department of ALM) as an example of cultural-heritage digitisation in spring 2023, 2024, 2025 (and continuing in 2026). For these students, a workshop on the project’s workflow was also held in autumn 2022 and autumn 2023.
The project has hosted a total of five student interns (three from History and two from ALM), who, within their advanced studies (7.5 or 15 credits), participated in the project and developed skills in machine reading and HTR.
Internationally, in spring terms 2022, 2023, 2024, and 2025, the project has contributed lectures and workshops to the course “Things That Matter: Materials & Culture in/for the Digital Age”, organized in collaboration between Uppsala University and the universities of Bern, Durham, Groningen, and Tübingen.
Outreach
The project has been communicated extensively through various channels. Updates and blog posts have been published on the website and on social media. A total of around 10 presentations have been given in relevant settings, including conferences, academic seminars, archives, museums, and other cultural heritage institutions. The project has also attracted media attention, appearing in newspapers, on television, on the radio, and in podcasts. An exhibition has been produced and shown twice in the Carolina Rediviva exhibition hall at UUL. In addition, the project has organised a lecture series with invited speakers—one per semester—at UUL.
The goal of making the collection and the project visible within the cultural-heritage sphere has thus been achieved.
The project has resulted in three publications:
• Mikael Alm & Johan Sjöberg, “Att bygga infrastruktur för forskning: Den Gustavianska samlingen vid Uppsala universitetsbibliotek,” in Henning Hansen & Peter Sjökvist (eds.), Från handskrift till digitalt kulturarv: Perspektiv på specialsamlingar, Uppsala, 2025, pp. 83–104
• Mikael Alm & Johan Sjöberg, “Dräkter för riddarspel och dramatik i Gustav III:s privata arkiv,” in Sofia Nestor (ed.), Teaterkungens kostymer, Stockholm, 2025, pp. 76–83
• Mikael Alm, “Det började med en skiss,” Historisk tidskrift (forthcoming 2026)
USE AND ACCESSIBILITY OF THE INFRASTRUCTURE
The project has created an infrastructure at UUL that can be applied to future projects. Establishing a workflow that spans the project’s components and units (conservation, image capture, HTR reading, accessibility, communication, and preservation) has been crucial for efficient work and avoiding bottlenecks. The outcome will benefit future projects at UUL.
The conservation and digital publication efforts are fully integrated into UUL’s operations. UUL is responsible for the collection and for Alvin. Alvin is publicly accessible.