Medieval Manuscripts Traced Through Family Tree

For every King Arthur or Roland, whose adventures readers can still enjoy today, another hero of ancient literature may have been lost forever. Before the printing press, texts were copied manually. This process introduced errors and innovations. Like mutations in the replication of DNA, these manuscript changes can be used to create evolutionary trees that philologists call stemmata. Since these trees are based on the extant copies, they do not reflect the full evolutionary history of texts and cannot account for the ones that are completely lost. Jean-Baptiste Camps and colleagues use a complexity science approach to estimate the amount of lost literature among chivalric narratives, beginning in the 12th century. Agent-based simulations suggest that up to 60% of texts and more than 95% of manuscripts may have been lost. The model reveals that the first few years after a text's creation are key: If few copies are made, the work is at high risk of extinction. The model also suggests that for most texts, no existing copies capture the original state of the work; all surviving texts are likely to be from a subsidiary branch of a given work's family tree. The oldest version of the Song of Roland is probably unknowable, for example. Seemingly random accidents or major historical contingencies, such as the Black Death, also likely led to the extinctions of texts. According to the authors, cultural heritage is fragile, and the model helps understand how randomness, historical contingencies, and human choices shaped the literature we inherit today.