To tackle a problem for scholars who study past cultures – the way numerous artefacts that would inform the cultural diversity of a time have been lost – researchers borrowed an “unseen species” model from ecology and used it to estimate the size of the original population of medieval European literature, and the nature of losses these cultural works sustained. Their results – which reveal surprising differences in which among these works persisted across Europe, pointing to island literatures as particularly resilient – highlight the utility of the authors’ innovative method for applications more broadly across the heritage sciences, from ancient coins to forgotten painters. Great amounts of literature that were a mainstay of culture in medieval Europe have not survived – due to library fires and to people disposing of them or recycling them in creative ways. In one famous case, as the authors will highlight in a video they’ll show at their AAAS Annual Meeting briefing, the remains of a medieval romance novel were used to strengthen a bishop’s headdress. As a result, scientists don’t know how representative the surviving literature is of what once existed, a phenomenon known as survivorship bias, which means they run the risk of underestimating the diversity of cultural production in medieval societies.
Here, to better account for medieval literature’s original scope, Mike Kestemont and colleagues took advantage of the way survivorship bias is also studied in ecology, which applies statistical methods to achieve bias correction. Kestemont and colleagues applied a well-known “unseen species” model called Chao1 developed by Anne Chao, one of the authors on this study, to provide robust estimates of how many species scientists are likely to have missed when out counting in the field. In their new research, Kestemont, Chao and colleagues assumed that literary works could be treated as species in ecology and that manuscript document copies of individual works could be treated as sightings of a species. They considered a work “lost” when none of the document copies that once preserved it survive any longer. This approach enabled them to estimate the size of the original population of works and of documents, respectively, as well as the losses that these cultural domains sustained, across six vernaculars (Dutch, French, Icelandic, Irish, English, and German). The authors note some of their observations have not been noticed before and challenge existing assumptions. The 3,648 medieval documents in the six vernaculars that are still observable today constitute a sample from a population that originally would have counted 40,614 specimens, they say. This translates to 9% survival rate. With respect to works, they estimate that about 68% survived, though they did observe considerable inter-vernacular variation, such as the relatively poorly surviving English works (38.6%). Further to the authors’ surprise, works for two of the more insular island cultures, Icelandic and Irish, were relatively intact, with survival rates of 77.3% and 81.0%, respectively. Additional analyses revealed something about these island literatures that has been overlooked in historical discussions of the survival of historic literature; namely, that they had a higher “evenness” – or more even distribution of copies for a given work – which helps create stability in the face of disasters like library fires. Evenness, say the authors, mirrors a special feature of islands as studied by ecologists: the endemic species richness is higher on islands than on the mainland. By contrast to what the authors saw for Icelandic and Irish works, the vaster medieval French literature had a low survival rate for works, which the authors attribute to many of its works being low in abundance (lacking evenness), rendering them more susceptible to immaterial loss. Kestemont and team say that, since there is nothing specific to ecology about the Chao1 method, their approach opens up a new way to study human cultures of the past.
Forgotten Books: The Application of Unseen Species Models to the Survival of Culture
Article Publication Date