First steps towards shared print collaboration in Sweden

Purpose – This paperreportsthe development in Sweden during the last few years onshared print cooperation. In June 2022, 30libraries signedaletter ofintent oncooperation onpreservation and withdrawal ofprintmaterial. By taking joint responsibility for long-term preservation, libraries aim to secure access as well as diversity and breadth in the collections. Design/methodology/approach – As a first practical step, the working group is conducting a national collection analysis using metadata in the national catalog LIBRIS. Findings – The paper presents the preliminary results of the collection analysis and discusses the next steps for the collaboration. Originality/value – The Swedish collaboration is one of many European and international collaborations on print archiving, and as such being of interest to a global library field.


Background
During the last few years, there has been an ongoing discussion on the need for collaboration and coordination of preservation and withdrawal of print material. A report from 2018 (Kungliga biblioteket, 2018) and an investigation from 2020 (Berglind et al., 2020) showed that there was both a need and an interest from libraries to work closer together on maintaining their print collections.
In October 2020, the National Library organized continued discussions to confirm the interest. In February 2021, they set up a steering committee and a working group for national collaboration. The steering committee is led by the National Library and contains representatives from both university libraries and special libraries (Samverkansgrupper. Bevarande och gallring, 2022).

Letter of intent
In June 2022 the steering committee agreed on a joint letter of intent that clarified the need for increased cooperation on preservation and withdrawal of print material. It also stated the need for libraries to take a joint responsibility for the long-term preservation of diversity and breadth in the print collections (Avsiktsf€ orklaring kring gallring och bevarande, 2022).
The letter states: "With this letter of intent, libraries want to clarify their shared responsibility and the need for national coordination to ensure the long-term supply of information in the form of print material." It also clarifies that the purpose is to secure access to print material "for the benefit of researchers, today and tomorrow, and in support of the democratic society". Libraries now need to find "forms for joint preservation and for coordinated and transparent withdrawal".
At first, the nine libraries in the steering committee signed the letter of intent, but only a few weeks later, more than thirty libraries had signed up for collaboration (Avsiktsf€ orklaring kring gallring och bevarande, 2022).

National collection analysis
In order to get a better overview of Swedish print collections, the working group has just begun conducting a national collection analysis. The analysis is based on metadata from the national catalog LIBRIS. All research libraries and many special libraries use LIBRIS.
A similar analysis had never been done before, so it was not clear what we would find. A useful source of inspiration and knowledge was a recent analysis from the UK called Strength in numbers (Malpas and Lavoie, 2016). That report looked at 36 research libraries in the UK and found that a relatively large amount of books only had a few copies. The conclusion was that scarcity is common, at least on a local scale.
It is important for the Swedish initiative to understand what is rare and what is common in a Swedish context. The collection analysis addresses the following questions: (1) How many records have few or only one holding?
(2) How many records have many holdings?
(3) How are collections distributed between library types? (national/research/special libraries) (4) How is the regional distribution?
It is also important to connect the Swedish results to a European and international context. We need to understand where Swedish libraries have unique material and when we can work together with international print management initiatives. The first step of the analysis is a pilot test, both regarding interest areas, data harvesting methods and results. Due to this, the following results must be considered highly preliminary and seen as examples of what is possible.
Preliminary results for print monographs For print monographs, there are 6m catalog records in LIBRIS. This includes monographs from all library types and in all languages. These records have about 20m holdings, on average just over three holdings per catalog record (see Figure 1).

Distribution of holdings for print monographs
The distribution of holdings between the catalog records is quite uneven. Almost half of the records (49%) only have one holding and three-quarters of the books have less than four holdings. LM Seven percent of the monographs have 10 or more holdings; this can be both popular student textbooks and children's literature (see Figure 2).

Language distribution
Of all the monographs in LIBRIS, less than one-third (28%) are in Swedish. Most of the books are in other languages, mostly English, German and French (see Figure 3).

Geographical distribution
By analyzing the data by the geographical location of the holding library, it is possible to get an understanding of the geographical distribution. Sweden is a country of large distances (1,572 km from north to south), so the geographical distribution of print material can be an important factor. The image shows an example with a subset of print monographs. The map shows a concentration of books to the Stockholm region in blue (National Library and Stockholm  The red dots are the collections of medium size and smaller libraries (see Figure 4).

Preliminary results for print journals
For print journals, there are 230,000 catalog records in LIBRIS. These records have about 700,000 holdings, on average just over three holdings per catalog record. Notably, even if the numbers are different, the ratio is the same as for print monographs (see Figure 5).

Distribution of holdings for print journals
Also for journals, a high number of titles (49%) have one holding. It is also worth noting that the holding for a journal title does not include the possible overlap for different years, so in reality there is probably an even lower holding overlap. However, it is interesting to notice that 19% of the titles have five holdings or more. Depending on the type of journal and the potential overlap, that could open up for opportunities for collaborative withdrawal.
Regarding the distribution of journal holdings, the ratio is very similar to the distribution for books. This result was a bit unexpected and will be examined more closely (see Figure 6).
Methodological and metadata problems Important aims of the first collection analysis were to learn more about LIBRIS as a data source and to find efficient methods for data analysis.
Data harvesting from LIBRIS was carried out with support from Ylva Sommerland at the National Library. The harvesting was done by using LIBRIS SPARQL Endpoint (Libris-OpenLink Virtuoso, 2022), and the data were analyzed using Tableau. However, the technical tools implied a number of limitations when working with large datasets and bibliographical data.
One major challenge with the metadata is that not all material is visible in LIBRIS. Collections from special libraries and extensive pre-1960 collections from the largest university libraries are not fully cataloged. This means there could be millions of records missing. Of course, this will also have huge implication on the number of holdings on some records. Several special libraries do not use LIBRIS for cataloging which means that their entire collections with many unique holdings can be missing.
The metadata also comprises a wide range of cataloging practices and cataloging rules over time. The analysis also shows difficulties related to certain subcategories, like separate reprints. This means there is inconsistency in how things will show up in the data sets. In summary, one important learning from this first collection analysis is the need for methods to analyze properly the data from LIBRIS. How to estimate the reliability of the results and how to manage the various sources of error? In order to conduct a more detailed collection analysis, the project also needs better software for data management and analysis of the large data sets.

Next steps
The preliminary results suggest that the future area of collaboration needs to focus on preservation rather than withdrawal. It seems scarcity is common both when it comes to monographs and periodicals in Sweden. Some withdrawal activities could probably be initiated on a smaller scale to find routines and establish the cooperation. One possible start could be coordinated withdrawal of duplicate copies of academic journals that also exist in electronic format. The working group will continue the collection analysis with a more defined dataset.
The Swedish collaboration needs to be more firmly established. Policies and priorities must be decided upon as well as organizational matters related to the long-term joint intent. Working methods, timing for scale-up and system support are other examples of questions that need to be solved in a near future.
Luckily enough, the European collaboration within the Epico Network has already provided the Swedish working group with contacts, practical information and inspiring