Rumors fly. Amidst a famine in 1789, tales of aristocratic conspiracies spread rapidly from town to town in rural France. The messages incited local groups to form militia to defend against brigands. When none emerged, the militia often turned on castles, land records, and other symbols of the aristocracy.

Riots flared up across much of the country between July 20 and August 6, ultimately leading to the end of the feudal system in France. The Great Fear's legacy is clear: more than two centuries later, France remains a democratic republic.

Yet, its precise origins were contentious. For decades, historians and sociologists debated precisely how rumors sparked the riots: were the protests born out of panic and outrage about unfounded information, or were they a rational response to the dire socio-political circumstances of the time?

In a recent study, economist Antoine Parent of the University of Paris teamed up with a multidisciplinary group of researchers to settle the debate. Instead of simply analyzing historical events, the group took an epidemiological approach, modeling the rumors as a kind of infection that spread from town to town. The model took into account historical records of the time, as well as road maps showing the paths that information might have traveled from town to town. The team found that while rumors spread across several towns, they only seeded riots in towns with specific risk factors, such as economic inflation. As with viruses, a combination of an "infectious agent" (the rumor) and a vulnerable population was necessary to cause a nearly nationwide outbreak of protests.

By doing so, the team discovered that the Great Fear and related protests were likely a result of specific socio-political risk factors, not panic alone. The work effectively combines a modern tool used to understand the spread of viruses and ideas and applies them to the poorly understood dynamics of a major historical event, says psychologist Sander van der Linden of the University of Cambridge in the United Kingdom, who was not involved with the work.

Previous studies have used epidemiological approaches to analyze how misinformation spreads in the modern world (see, for example, here, here, and here) or how ideas go viral on social media. So approaching the origins of the Great Fear in this way was "pretty obvious," says study coauthor Caterina La Porta, a pathologist at the University of Milan in Italy. "The less obvious thing was the data -- we didn't have a large dataset."

To gather the data needed to develop a model, the researchers turned to French historian Georges Lefebvre's 1932 book The Great Fear of 1789: Rural Panic in Revolutionary France, where rumors spread and when they spread. Sometimes it was just a date, but in other instances, the book reported the precise hour when a rumor landed in a town. The team also included maps of the road network and postal stations from the time to discern the paths rumors may have followed.

Then, they overlaid this transmission of gossip with "risk factors" that might have made a town more or less likely to be "infected." These included, for example, literacy rates, local land records of feudal ownership, and economic conditions such as the price of wheat. The epidemiological model then helped the team identify which risk factors were linked to the local revolts. For instance, they found that towns with higher prices of wheat were more likely to have experienced the Great Fear. And riots were more likely to have occurred in towns with written records of feudal land rights, where peasants could storm castles and burn land records that supported aristocratic privilege.

"Putting all these factors together, we found they aligned toward political grievance as a reason for fear," says lead author Stefano Zapperi, a physicist at the University of Milan. "The idea that it was just irrational riots based on emotions doesn't stand."

The work supports the argument that protests during the Great Fear were usually rooted in rational interests, says Jack Goldstone, a sociologist at George Mason University in Virginia, who was not involved in the study but authored an accompanying commentary. "This analysis is definitive in terms of laying to rest the idea that the bulk of the action was a panic by ill-informed individuals," Goldstone says. "It shows quite definitively that it was a rational response by communities."

However, van der Linden notes that it's not possible to exclude a role for emotions completely. Studies of modern data suggest that outrage and negative emotions are a significant predictor of what information goes viral on the Internet. Although social and political factors in towns point to a rational explanation, "at the individual level, there's probably some role that emotions played." The historical data the researchers used aren't granular enough to elucidate and gauge the psychological mechanisms of information spread or to determine language used to communicate rumors, van der Linden says. "It is a limitation for understanding exactly what goes on."

He does believe that, in some ways, the debate over the origins of the Great Fear echoes present-day discussions about why people spread misinformation. Do they think they're being rational? Are their emotions largely driving the spread? "It could be both," says van der Linden.

Goldstone adds that the idea of a rational purpose underlying protests, as seen during the Great Fear, may not extend to all mass uprisings. Other protests or movements in Europe at the time may have been driven by famines, religion, or other factors. He thinks the epidemiological approach developed in this study could help illuminate the drivers of other historical events.

According to van der Linden, such models could eventually also help predict protests or outbreaks in response to the spread of misinformation in the modern world. "I think that you could create population-level models that not only use knowledge about how rumors diffuse," he says, "but then also, what we can do about it?"