A New Tool for Describing Provenance Images CERL’s Provenance Digital Archive

Following the rise of the study of the history of the book in the eighties and nineties, provenance studies have become an important component in the research of social and cultural historians. This development was noticed and embraced by the Consortium of European Research Libraries (CERL). The Consortium incorporated results of provenance research in existing resources and new ones were developed. The latest development is the CERL Provenance Digital Archive, an international online database to describe images of provenance marks.


Introduction
Suppose, you are researching the provenance of an incunable and you find the bookplate that is shown in fig. 3? How would you find out who owned the book in former times? After a period of studying book production and the dissemination of the book through publishers and book shops, questions of readership became a main interest of social and cultural historians dealing with literacy and the spread of ideas. At the same time, a separate developing concern about the security of library collections has widened the inter-est in recording provenances of all sorts, ancient and modern. These developments stimulated The Consortium of European Research libraries (CERL) 1 to incorporate provenance research among its goals.

CERL and Provenance
Fifteen years ago, David Pearson, author of the handbook for provenance research 2 wrote in the proceedings of one of CERL's conferences: "The idea of creating a database which links biographical and bibliographical information about early book owners with their distinctive marks of ownership seems an obvious possibility and one which could be useful to cataloguers and researchers alike once it developed a critical mass". 3 Since then, CERL developed many resources that provide an environment for the recording of provenance data for both manuscripts and books of the hand-press period. CERL's tools and databases register provenance data in various levels of detail. First of all there is the Heritage of the Printed Book Database (HPB). This is a steadily growing collection of files of catalogue records from major European and North American research libraries covering items of European printing of the hand-press period (c. 1455-c. 1830) integrated into one searchable environment. Increasing numbers of records in this database record provenance information that is made searchable in the HPB. Then there is the CERL Thesaurus. This resource contains multilingual information on names and variant names of persons and places pertaining to books of the hand-press period. Provenance Names are also included in this database, they are marked as such and again retrievable under this particular aspect. The Material Evidence in Incunabula (MEI) is specifically designed to record and search material aspects of 15th-century printed books by capturing ownership and other provenance evidence such as bindings, manuscript annotations, book prices and bookplates. The same model has been applied to post-1500 printed books in a project generated database currently named Patrimonit. All resources mentioned until now are text based. But provenance often manifests itself physically: in an ex-libris, a signature, an inscription, a stamp, etc.; therefore, the association of a provenance description with an image was wished for. For a long time CERL has strived to develop a resource that could include descriptions and images. A first step was achieved with the Can You Help section at CERL's website since 2009, managed by David Shaw. Users could send provenance queries (including images) to David Shaw and post replies to other people's requests. As the number of questions grew, this approach became unmanageable and poorly usable. This stimulated the search to develop a resource where images could be uploaded and described: the Provenance Digital Archive (PDA), a database where images of provenance marks can be stored and searched. This tool should offer researchers, users of and contributors to certain CERL services such as Can You Help, MEI and Patrimonit, the possibility to upload images, describe the content and connect them to other CERL resources (and other provenance databases).

3
The Development of the Provenance Digital Archive CERL's Provenance Working Group formulated a set of user scenarios to make sure that the tool CERL was going to develop would be useful for our target group of cataloguers, bibliographers and researchers. In short, we wanted the system to be a central place for provenance images where libraries and scholars could store photographs they took during their cataloguing activities and research and that helps them to identify the unknown provenance marks they encountered. These wishes led to the following list of requirements: 1. an intuitive and simple description model that includes all necessary elements but is not too detailed, so participants can easily describe and retrieve provenance marks; 2. possibility of uploading single images with basic metadata; 3. possibility of batch uploading of images with basic metadata provided by the uploader; 4. editing facilities for registered users; they should be able to edit their own records and comment on records of other users within the CERL PDA; dialogue visible to all; 5. possibility to add the identification to unidentified provenance marks; 6. integration of current Can You Help facilities in the new resource; facility for alerting new questions to identify provenance marks to registered users; 7. possibility to download images; therefore the images provided should be free of copyright; 8. possibility of replacing images with better ones; 9. possibility of adding more than one image for one provenance mark; 10. possibility of adding different provenance marks pertaining to one source; 11. facility to link to the existing CERL Thesaurus records and add new names in the CT where necessary; 12. adding extra sources when a provenance mark occurs in more than one source; 13. the CERL PDA must be open access.
In Arkyes, CERL found a partner that would be able to embed the desired functionality in an existing and rich infrastructure. Arkyves is based in Leiden and related to Brill Publishers. 5 It is a "collection of image collections" from museums, libraries and research institutes worldwide, among them the Rijksmuseum in Amsterdam, the Netherlands Institute for Art History (RKD), and the Herzog August Bibliothek in Wolfenbüttel. Emblems, book illustrations, printed initials, bookplates and so forth are brought together in one large database, each description sharing one feature: they all use the same standard subject classification for cultural visual content: Iconclass. This is a classification system designed to describe the contents of an image, using letters and numbers instead of language. 6 4 The CERL PDA The Arkyves team, now part of Brill, developed the CERL PDA as a separate, yet integrated, section in the Arkyves collection. It can be approached via the Arkyves website by clicking the "CERL Provenance Digital Archive" bar in the website of Arkyves. To upload and describe images, one has to log in with a Google, Facebook or Orcid account; searching the CERL PDA can be done without logging in. Some 140 items are described as a test, and these already give an impression of the many possibilities offered by the CERL PDA. One can find ex-libris, bookbindings, inscriptions, armorial images etc. As with any database, its value as a research tool will increase with the growth of its contents.
6 URL http://www.iconclass.nl/home. Iconclass consists of a hierarchically ordered collection of definitions of objects, people, events and abstract ideas that represent the subject of an image. There are 9 main divisions, further specified through subdivisions. For example, the main division 'Nature' (2) is further divided in 'earth' (25), holding the subdivision animals (25F), holding the subdivision mammals (25F2) and so on. So every image can be captured in numbers and letters. The classification system is thus language independent. For every provenance mark, a short description is given: the source (i.e. in which book the provenance mark was found), the institution that holds the book, the measurements of the provenance mark, the type, the technique, the time period and the location of the mark in the book. And of course, the most important characteristic: the owner indicated by the provenance mark, i.e. the person or institution that connected the provenance mark to the book, the former owner of the book. Elements as the source and the names of former owners are linked to other CERL resources such as MEI, ISTC, HPB and the CERL Thesaurus; from these resources they can be linked to the PDA as well, now or in the future. Furthermore an Iconclass code is added. Every element in the description is indexed so that all elements in the description can be searched, separately or in combination.
The last element of the description, the Iconclass code, is added to make subject retrieval more adequate. This Iconclass code indexes the subject of the provenance marks at different levels. An example will explain this. To the bookplate [ fig. 2] of the Dutch physician Cornelis Hendrik à Roy, 7 the following Iconclass codes were connected, describing on the one hand the kind of object (ex-libris, mark of ownership) and on the other the elements of the image (horse, helmet, foliage): • 49M8 ex-libris • 49L27 mark of ownership • 46C13141 horse • 45C221 helmet • 48AA9831 foliage, tendrils, branches ~ ornament -AA -stylized This is a relatively elaborate Iconclass subject indexing; a more basic description will suffice for the record to be accepted in the database. CERL also strives to collaborate with other provenance resources and aspires to support bulk uploads, so that the pool of data will rapidly be increased and expanded. Then, more results can be offered to researchers and cataloguers searching for the former owner associated with a certain provenance mark. Now that the database is developed, users can start uploading and describing the images they collect during their research. For this, a set of guidelines is available. 8

5
Benefits of the CERL PDA The development of the CERL PDA was finished in September 2019 after intensive testing. Now CERL expects that almost all of the requirements defined above are met. During the testing phase some 150 provenance marks have been added, mostly originating from the Koninklijke Bibliotheek (KB), The Hague. These are provenance marks that were found while contributing to CERL's Material Evidence in Incunabula database, so they were found in the incunabula of the KB. Although the corpus at this moment is still small, users can already consult the database when they find a provenance mark that they do not know. If, for example, someone is cataloguing a book with a stamp VF in the middle of a wreath (garland), this person can search the PDA with the words "VF" and "wreath" and will find an  Another example: if a researcher encounters books with a specific notation of prices and numbers, he/she can also search the CERL PDA with specific words such as "inscription", "price" or "circle". The user will then find an image [ fig. 4] that might have the same composition as he found in the books he is describing and will then know that this book once belonged to the Leiden lawyer Joost Romswinckel (1745Romswinckel ( -1824. 10 And of course, the usability and benefits of the CERL PDA will further increase when more provenance marks are inserted. Another gain of the CERL PDA is that it can be used as Can You Help resource. In the description a box can be ticked if the user would like to share the image with other users hoping that other people can help to identify the mark. Someone can upload an image he wants to identify, tick this box and then it will be shown via an RSS feed to researchers that indicated that they are willing to help to identify the persons or institutions that used a certain image. These names then can be connected to the descriptions in the CERL PDA. 10 Image of the notation of purchase information of Joost Romswinckel: https://www. arkyves.org/r/view/cerlpda_daad0/him_CERLPDA. The CERL PDA is developed as a section of Arkyves. From this follows that it is part of a very large collection of images from various European and American institutions. In the context of Arkyves, more images of provenance can be found deriving from other collections. Here, it should be mentioned that, while the CERL PDA is a free resource, Arkyves as a whole is a subscription database held by Brill, Leiden. This means that when a single user searches the CERL PDA, he/she will find the images described in this specific section, whereas a subscriber to the full Arkyves database would be presented with additional provenance marks. If, for example, a subscriber to Arkyves wants to identify a bookplate that is not yet described in the PDA, he might find it in the rest of the Arkyves collection [ fig. 5]. This is true, for example, for a bookplate with a snake, a bird and branches that is held in the Herzog August Bibliothek (Wolfenbüttel) as part of their Virtual Print Cabinet. Then the user would find that this was the bookplate of Christian Johann Lange. Moreover, even if no similar image is found, related images may point a user in which direction he/she should continue the research. Still, to be clear, a user does not have to subscribe to Arkyves to use the CERL PDA. This is an open resource of CERL. On its own, the CERL PDA will grow into a very useful book historical tool, becoming more and more beneficial as it expands.
In the past years, CERL became more and more connected to provenance research through the development of the Material Evidence in Incunabula and Patrimonit databases. The need for an international provenance image database that could be used by anyone studying manuscripts and books from the hand-press period and beyond was felt and CERL undertook steps to develop a resource for this. The aim was that users could upload their images and describe them in an intuitive way and others should be able to search this image database and find the information they are looking for. Via the CERL PDA, the Consortium hopes to support the work of many researchers and cataloguers that will fill and use this resource and will contribute to building an international, open, frequently-used online tool for the benefit of the study of books and their history.