Written by Connor Zahler:
Data Detectives is a NEW series from Putting Methods to the Madness based on the assignment of the same name from QMSS 301. Student writers find a statistic or talking point and do “detective” work by tracking down its sources, how it has spread, and what impact it has today. In this inaugural post, Connor Moore tackles a historical question: how many witches were burned during the European witch hunts?
Throughout the Early Modern people, witch crazes swept through Europe and parts of Colonial America. Many people (mostly women) were accused of maleficium: harmful magic meant to hurt other people. Rumors swirled of the abduction and sacrifice of newborns, mass gatherings of thousands of witches with the devil himself present, innocent cows being robbed of their milk via sticking an ax in a wall, and even more. These events have captured the minds of observers, both those contemporary to the witch hunts and those after. One question has dogged them throughout time: how many people actually died?
This question is made both easier and more difficult due to the records we have. On one hand, many witch hunts were legal procedures that left behind interrogation records, court decisions, and other documentary evidence. We also have multitudes of personal testaments left behind by witch hunters, judges, and regular people caught up in the trials (usually as observers, but sometimes as victims). Unfortunately, we are also missing many records. Records can be lost due to purposeful destruction, simple misplacement, or any other number of reasons after hundreds of years. There’s also been concern about the possible fabrication of records. In the past few decades, it has been revealed that the events initially considered the first witch hunts did not occur (but that’s a story for another day).
Despite the evidentiary difficulties, many have tried their hand at an estimate. One of the first was Voltaire, the French intellectual and writer. Writing in the mid-1700s, even before the last execution for witchcraft on the European continent, he estimated that several hundred thousand people had been executed. As we will see later, this isn’t bad, especially given that he lacked access to almost any demographic evidence. Unfortunately, Voltaire’s moderate estimate was soon overshadowed by another estimate made in response to it: one made by Gottfried Christian Voigt.
Today, Voigt is known almost entirely for being wrong in a truly extraordinary fashion. Incensed by how low Voltaire’s estimate is, he did some questionable math. He started with one region of Germany, which recorded 133 executions from a population of 11-12,000 across a century. He then applied this ratio to all of Christendom to get 858,454 per century. For eleven centuries (600-1700 C.E.), this amounts to a grand total of 9,442,994 people executed for witchcraft. To make a long story short, he was incredibly wrong. Modern estimates put the number somewhere in the range of 30-70,000, making Voigt off by about 9,370,000. How did he go so wrong?
In any statistics-based class (including QMSS), you quickly learn about the dangers of overgeneralizing. For example, let’s say I make a study that says 90% of students at U of M read my posts on the student blog and enjoy them. This would mean that somewhere around 27,000 people consume my content, which is a bold claim to make, and pretty clearly incorrect (for now!). How could I have reached that percentage? Well, it turns out I asked ten people in my QMSS 301 class, nine of whom were my friends. I’ve taken a sample that isn’t representative of any wider population and applied it to that population, resulting in a wildly inaccurate estimate. Voigt did the same thing: he took one hotspot of witch persecution and applied it to an entire continent.
There’s also the question of time. In history, we are often faced with the problem of presentism: the idea of applying the way things are today (whether it be morals, intellectual concepts, or any other idea) to a completely different historical context. Voigt doesn’t exactly do presentism, but he makes a similar mistake. For one, witch hunts did not occur for eleven centuries. A roughly accurate time range would be 1450-1750, with the vast majority of persecutions from 1550-1650. This connects to a second issue: the ferocity of witch hunts varied greatly across time and place. There were cities that recorded a handful in a hundred years and cities that recorded a hundred in a handful of years. When you homogenize vastly different times and places, you end up with statistics that are so incorrect as to be not only useless, but actively dangerous.
In the next part of this series, I’ll tackle the fallout of Voigt’s massive overestimation. His sloppy statistics are part of a story that encompasses Christian sectarianism, the foundation of a world religion, the Second World War, and the American feminist moment.