Digital Humanities Scholars Map a Forgotten Art Form Hidden Inside Classic Novels

Books that include maps have always felt a bit like literary treasure, but new research from Cornell University shows just how rare they really are. Scholars working in the field of digital humanities have now used large-scale computational tools to uncover and study maps hidden inside novels from the 19th and early 20th centuries. What they found challenges long-held assumptions about literature, genre, and how readers have historically navigated fictional worlds.

At the center of this research is a simple but surprisingly difficult question: how often do novels actually include maps? Until now, no one had a clear answer. The scope of published fiction was simply too large for traditional literary analysis to handle. By combining artificial intelligence with literary scholarship, the Cornell research team finally managed to answer it.

Mining Nearly 100,000 Novels for Maps

The study was conducted by researchers from the Cornell Ann S. Bowers College of Computing and Information Science, led by doctoral student Axel Bax, with guidance from associate professor Matthew Wilkens. The team examined nearly 100,000 digitized novels published between 1800 and 1928, drawing from more than 32 million scanned pages housed in the HathiTrust Digital Library.

Using a batch of neural network models trained for image recognition, the researchers built a system capable of identifying maps embedded in books. These models scanned page images and distinguished maps from other illustrations such as tables, diagrams, or decorative artwork. Human reviewers then verified the results to ensure accuracy.

The findings were striking. Only 1.7% of novels from this massive dataset contained at least one map. In other words, books with maps really are literary rarities, much closer to buried treasure than most readers might expect.

Where Maps Appear in Novels

When maps do appear in fiction, they tend to show up in predictable places. The researchers found that most maps were located at the very beginning or very end of novels, often as part of the front matter or appendices. Maps embedded within the main body of the text were far less common.

This placement suggests that maps were intended as orientation tools, helping readers understand the setting before diving into the story, or as reference materials to revisit after finishing the book.

Fictional Worlds Versus Real Geography

One of the most interesting discoveries was that 25% of the maps depicted fictional locations, while the remaining majority represented real places. This finding matters because it expands how scholars can study character movement in literature.

Previously, computational studies of literary geography were limited to real-world locations that could be mapped using historical records. By identifying fictional maps, this new system allows researchers to analyze how characters move through imaginary spaces as well.

This opens the door to large-scale comparisons between novels set in real places and those set in invented worlds, something that had not been possible before.

Genre Results That Defied Expectations

Many people would assume that fantasy or science fiction novels are the most likely to contain maps. After all, modern epic fantasy often relies heavily on detailed cartography. However, the data told a different story.

The genres most likely to include maps were military fiction and detective fiction, not fantasy or science fiction. This surprised even the researchers. Military novels often included maps to clarify troop movements, battlefields, or campaign routes, while detective fiction sometimes used maps to explain crimes, neighborhoods, or pursuit paths.

This discovery challenges modern assumptions shaped by contemporary fantasy publishing and shows that historical genre conventions were very different from today’s.

A Hidden Problem in Library Records

Another major finding involved library metadata. The researchers compared their results with existing MARC records, the standard machine-readable format libraries use to catalog books. They found that more than half of the novels containing maps were not labeled as such in library records.

This means countless literary maps have effectively been invisible to scholars searching catalogs. The team’s system could help libraries identify and correct these omissions, improving access for researchers and readers alike.

Why This Matters for Literary Scholarship

Before this study, questions like “How often do maps appear in literature?” or “How has map usage changed over time?” were nearly impossible to answer. Literary scholars had no realistic way to manually scan tens of thousands of books.

By using computational tools, the Cornell team demonstrated how digital humanities can uncover cultural patterns that close reading alone cannot reveal. Their work shows how literature reflects historical attitudes toward space, movement, and geography across entire centuries.

The researchers also found that novels containing maps tend to use more spatial language overall, suggesting a deeper narrative focus on movement, distance, and place.

Understanding Character Movement at Scale

One of the long-term goals of this project is to analyze how far characters travel within stories. With maps now identified in both real and fictional settings, scholars can begin asking whether characters in imaginary worlds move differently than those grounded in real geography.

This type of analysis could reshape how we understand adventure narratives, travel writing, colonial literature, and genre fiction, all through the lens of spatial movement.

The Broader Role of Digital Humanities

This project is a clear example of how digital humanities blends computing and the humanities to tackle questions that once seemed unanswerable. By working at scale, scholars can move beyond anecdotal evidence and test long-standing beliefs about literature using data.

As more libraries digitize their collections, methods like these will become increasingly important. They allow scholars to revisit familiar texts while also discovering patterns hidden across thousands of forgotten books.

Looking Ahead

Future research will build on this dataset to compare fictional and real spaces in greater detail. With a sufficiently large sample size, researchers hope to determine whether characters behave differently depending on whether their world exists on a real map or an imagined one.

What started as a search for a rare literary feature has turned into a powerful tool for understanding how stories organize space, movement, and meaning across time.

Research paper: https://ecommons.cornell.edu/items/6d470f13-bac5-4b0e-94f2-4eac93d677cb