Katie McGettigan is a lecturer in American literature at Royal Holloway, University of London. Her first book, Herman Melville: Modernity and the Material Text, is forthcoming from the University Press of New England, and she is working on a study of the publication of American literature in England, 1830-1860, funded by the Leverhulme Trust. Katie attended our Digital Antiquarian conference and workshop in the summer of 2015.
Of all the things that I gained from attending the AAS’s Digital Antiquarian workshop in 2015, a fascination with MAchine Readable Catalog (MARC) records was definitely the most unexpected. The workshop’s sessions on digging into and manipulating MARC records sowed the seeds of a research project that will, I hope, shed light on a neglected publishing genre of the nineteenth century.
Before attending the workshop, I’d given little thought to how a book is cataloged. I learnt that a cataloger enters information about the book in MARC format, breaking down its different attributes into individual fields: for example, the 100 field contains information about the author, and the 260 field contains publication information. Kathleen Haley, the AAS systems librarian, showed us that by downloading MARC records from the AAS catalog and loading them into MARCEdit, we could conduct more complex searches than were possible through the online catalog alone. She also showed us how these records could be exported into a CSV spreadsheet file for data analysis (there’s a great tutorial video on Past is Present if you want to try this yourself).
The workshop set me thinking about problems I was having in researching publishers’ series as part of a project on the publication of American literature in Victorian Britain. Publishers’ series consist of volumes by different authors, issued by a single publisher under a general series title, often in a uniform format. It was a popular form of cheap publishing in the nineteenth century, and famous British and American examples include George Routledge’s “Railway Library,” published from 1849, and Harper Brothers’ “Family Library,” which grew to two hundred titles. British publishers often turned to American titles to fill their series because their U.S. copyrights did not apply overseas. George Slater, a London publisher who specialized in cheap print, reprinted Margaret Fuller’s Woman in the Nineteenth Century in his “Shilling Series” in 1850, following it with a volume on needlework.
While there were thousands of series published in the nineteenth century, trying to find which titles they included can be tricky. There are detailed and extensive catalogs of fiction series by successful publishers like Routledge, but these don’t exists for smaller publishers like Slater, and non-fiction series. Searching library catalogs for these smaller series is challenging because many catalogs don’t allow searching by series title – only by the title of the text or the author (AAS helpfully catalogs publishers’ series separately). The Workshop showed me that downloading and searching MARC records directly could help me to locate more series titles, but I also wondered whether MARC records could be used as a dataset to enhance our picture of publishing in the nineteenth century.
Returning to the UK, I began to collaborate with Dr. Paul Rooney, an historian of British and Irish publishing at the National University of Ireland, Galway, on how we might use library metadata to fill gaps in our knowledge of nineteenth-century series. Helpfully for us, MARC records contain a field for “Series Statement” in field 490; if a title is published as part of a series and the cataloger includes that information, it should have an entry here. We thought that pulling MARC records of titles with a 490 entry, published between 1800 and 1900, would create an alternative dataset of publishers’ series that would allow us to explore trends at a much larger scale than individual series or publishers. We could chart, for example, whether certain authors tended to cluster together in series and how titles moved through different genres of series as the century progressed.
We received help from a Data Driven Discovery Grant from Nottingham University, and from the British Library Labs team, who gave us access to the catalog records for the library’s 1.8 million nineteenth-century titles in MARCXML format. We put these records into a MYSQL database, which we then queried for all titles published in London with an entry in the 490 field. Because this also gave us other types of publications in series (law reports, sheet music, parliamentary papers), we manually extracted these using data cleaning software called Open Refine, which also allowed us to regularize variations in, for example, the names of publishing houses.
Comparing our dataset to what we knew of the Victorian publishing landscape revealed a lot of publications missing. We’d been relying on the cataloging being consistent and complete, but information about the series title might not have been included when the book was first cataloged, or might have been entered in a different part of the MARC record when catalog cards were digitized. To compensate for this, we revised our query to search for known titles in other parts of the MARC record (series information is often in the 500 “notes” field), and we’re working on refining our searches further. But there will also be books that aren’t in the British Library, which is why we hope to collaborate with other libraries and introduce their records into our dataset. But our research thus far is a reminder that working with library records reveals as much about collecting and cataloging priorities as publication history.
Nevertheless, we have started thinking about how to analyse the data we have produced. We are experimenting with network graphs of authors whose works were issued by the same publisher, or in the same series; this works best for fiction titles, which were often reprinted in several series. From the early graphs that we’ve produced, distinct groupings of authors are emerging. Eighteenth-century authors, whose works were out of copyright, tend to cluster together, as do authors of contemporary fiction. This suggests that publishers specialized in particular types of fiction series, offering either classic titles or the latest productions. Some authors, like Daniel Defoe, connect both groupings, pointing to the appearance of Robinson Crusoe in series aimed at many different audiences. These results are, of course, skewed by cataloging priorities. Our database queries returned large numbers of titles by writers like Charles Dickens and Thomas Hardy, whose canonical status means that more care is likely to be taken with the cataloging of their work; this might lead to them seeming more well-connected in author networks than they actually were. As we add more library records, we expect our groupings of authors to shift.
As we grow our dataset, we hope that our digital approach will further illuminate how Victorian publishers constructed their series, and how this publication genre evolved across the century. As I learned from my week with AAS catalogers, we must also remain mindful of the decades of work that went into creating this big data for the nineteenth-century book. Like any work that spans time and place, inconsistencies are inevitable, and we must account for them as we analyse and interpret our findings.