Werewolves, and the judge who called it an
illusion.

The boundary

Three meanings of real, written down before any source was read.

As in the vampire run, the question was split before any searching began: were the trials real events, were natural things misread as werewolves, and did anyone actually change shape?

The period runs from classical myth through the trial era, roughly the 1520s to 1692, to the Beast of Gévaudan in 1764–67, and ends at The Wolf Man in 1941, which is in scope only to show where the modern rules came from. The places are the trial clusters: Franche-Comté and the Jura, the Rhineland around Bedburg, and Livonia, in today’s Estonia and Latvia.

Fifteen subquestions were written first: what the confessions actually contain and how they were taken, whether werewolf trials differed from witch trials in the same courts, whether accused werewolves were mostly men, whether clinical lycanthropy can be diagnosed four hundred years late without anachronism, and whether rabies, ergot, porphyria or excess body hair fit any recorded case.

One ambiguity was written down at the start: the Beast of Gévaudan may not count as a werewolf case at all, because nobody was ever tried for it. That turned out to be the point.

The question was not whether werewolves exist. It was which kind of real the record can support.

Competing explanations

Four explanations, and the literal one came last.

Each explanation was written down with the observation that would destroy it. Eighty-seven passages were then scored against all four.

Fusion (H1), an old folk belief absorbed into the witch-hunting machine, collected forty-four supporting passages and five against. Misread natural causes (H2) split almost evenly, twenty for and twenty-one against, because the wolves and the killings held up and the diseases did not. Opportunism (H3), the werewolf charge as a convenient weapon in local and political fights, came out slightly positive.

Literal reality (H4), that shapeshifting was real or that the courts prosecuted it as real, drew six supporting passages and thirty-seven against. Every one of the six is about someone believing in werewolves, not about anything turning into a wolf.

The second round of searching added three new source families, and it pushed the counts further towards fusion and further against the literal reading.

Every passage that leaned towards the literal werewolf was about belief, not about a body.

Where the belief came from

The wolf-man was ancient. The Church first called him a dream.

The run found the werewolf long before any court did. What changed in the trial era was not the belief but the official attitude to it.

The oldest writing on lycanthropy the run found is the Greek myth of Lycaon, king of Arcadia, whom Zeus turns into a wolf. In Ovid’s telling his clothes become shaggy hair, his arms become legs, and he keeps his greed for blood. Scandinavian warriors, the ulfheðnar, ate wolves and wore wolf fur into battle. The belief was in place long before any demonology.

The medieval Church was sceptical of it. The Canon Episcopi held that such transformations happen in the mind as dream visions, not in the body, and that anyone who believes a person can become another kind of creature is worse in faith than an infidel. Burchard of Worms added: “and worse than a pagan.” Believing in the transformation was the heresy.

That had to be argued away before the trials could start. The run’s source contrasts the canon with the 1484 bull Summis desiderantes, and notes that the authors of the Malleus Maleficarum, the Hammer of Witches, knew the old canon stood in their way. Even after that, as chapter six shows, the demonologists never quite accepted that a body changes shape.

In the medieval canon, believing in werewolves was the heresy.

The trials

One case has witnesses. The other has only a confession on the rack.

The two famous werewolves are usually told as one story. The run found their evidence is opposite in kind, and that what is true of one cannot be carried over to the other.

Gilles Garnier, a hermit near Dole, was burned on 18 January 1574 for “crimes of lycanthropy and witchcraft.” His case carries evidence that did not come from his confession. More than fifty witnesses deposed that children had been attacked and killed in the fields and vineyards. A boy of twelve or thirteen was found bitten and torn open. And before any arrest, workers travelling at dusk came on what looked like a wolf, which some recognised as the hermit, with the body of a dead child. Then came the confession under torture: a spectre in the forest, and an ointment that turned him into a wolf.

Peter Stumpp, executed near Cologne in 1589, has none of that. No interrogation transcript or court record survives; almost everything known comes from a sixteen-page London pamphlet of 1590, translated from a German print that is lost. He confessed after being stretched on the rack, under threat of more: a magic belt from the Devil, fourteen children and two pregnant women eaten, incest, and a succubus. The proof offered was that a wolf had lost its left forepaw, and he had the same injury.

The run’s conclusion is that confession content tracks the interrogator’s questionnaire, not any disease or predation signature. The more vivid the confession, the less it proves.

The more vivid the confession, the less it proves.

How a werewolf became a witch

The accused never mentioned Satan. The court added him.

The strongest finding in the run is about procedure, not belief, and it has the right order in time: the neighbours’ werewolf accusation comes first, and the Devil arrives later, in the interrogation room.

The clearest description comes from the Baltic. Peasants there believed in harmful magic and in werewolves, but, worshipping outside the Christian frame, not in Satan or pacts with him. When they accused someone of doing damage in the shape of a wolf, the authorities read it as witchcraft. Through leading questions and torture, confessions about werewolves were adjusted into confessions about Satan, and the accused were executed as witches.

That fits the rest of the record. Werewolf charges appear in only a small fraction of witchcraft trials, as an add-on available inside existing machinery. Garnier was convicted of “lycanthropy and witchcraft” together. And the scapegoat reading has support at Bedburg: the Catholic faction had just taken control, Stumpp had reputedly converted to Protestantism, and a brutal trial would warn Protestants off rebellion.

The weak point is that most of this comes from one place. The run traced its seven supporting passages to a single independent origin, and one of its stress tests names what would break the finding: records in which magistrates started with the Devil before any neighbour had said “werewolf.”

The neighbours said werewolf. The interrogators wrote down the Devil.

What the judges believed

The judge who tried werewolves wrote that no one becomes a wolf.

Literal shapeshifting lost twice: once to the evidence, and once to the men who ran the trials.

The one primary text the run read is a chapter of An Examen of Witches, the English translation of Henri Boguet’s treatise; Boguet was a judge in Franche-Comté who tried accused witches and werewolves himself. He is blunt: “it has always been my opinion that Lycanthropy is an illusion, and that the metamorphosis of a man into a beast is impossible.” His theory was that the Devil confuses the witch’s imagination, so that the witch really runs about killing while believing he is a wolf, or that Satan does the killing himself in wolf shape while the witch sleeps behind a bush.

He used the trial evidence against transformation himself. Witnesses had recognised the hands and toes of two accused women, Perrenette Gandillon and Clauda Gaillard. “Are they not evidence,” he asked, “that they were not really transformed into wolves?” He added that a Church council had declared believers in metamorphosis to be infidels.

So the legal theory was never that a body changed shape. It was that a deluded human being really committed the killings. That is why a conviction cannot tell you which explanation produced the defendant: the same theory fits a serial killer, a scapegoat and a demonological framing equally well.

“Lycanthropy is an illusion, and the metamorphosis of a man into a beast is impossible.”

The popular theories

Real wolves killed thousands. Real diseases look wolf-like. Neither shows up in a trial.

The natural explanations split cleanly. Predation was real and large. The medical theories establish only that a mechanism is possible.

Wolves really did kill people. One estimate puts French wolf-attack deaths at as many as 9,000 between the late sixteenth and early nineteenth centuries. In Gévaudan the victims were almost all women and children, the first a fourteen-year-old girl watching her sheep, many with throats torn out. Officials at the time described something bigger than a wolf, hyena-like or lion-like; popular retelling added walking on hind legs, bullet-proof hide and returning from the dead. The historian the source quotes reads it as a serious wolf infestation rather than one beast. A farmer finally killed the creature with an ordinary bullet, and no one was tried. Predation was a background condition, not a cause of trials.

The diseases are real too. A systematic review found 43 cases of clinical lycanthropy, or its dog equivalent, reported between 1852 and 2020, mostly in schizophrenia, psychotic depression and mania. Congenital hypertrichosis and Günther’s porphyria cause excess hair, scarring and red-stained teeth. But the review rates its own evidence as low, built on case reports, and none of these conditions appears in any trial record the run saw.

A peer-reviewed history of medicine adds that the rabies, porphyria and ergot theories were mostly published in medical journals by history enthusiasts, and rebuts the ergot and witches’-ointment theories. Physicians at the time offered melancholy as the explanation, while knowing the classical melancholic lycanthrope was harmless rather than murderous.

A disease that makes a person look like a wolf is not a disease that appears in a trial record.

Trying to break it

Eight searches for the other side, and six that came back.

After the first pass, the run went looking for the evidence that would hurt each explanation most. This time most of the counter-searches found something.

Both searches against fusion returned the same overview of werewolf witch trials, and it supplied the mechanism and the exception to it. Thiess of Kaltenbrun, in his eighties, volunteered in a church-robbery case where it was irrelevant that he had been a werewolf. Werewolves, he said, were God’s hounds who fought witches in Hell to bring back stolen grain. The court could not get a pact with the Devil out of him. Where the peasants had no Satan and the court got no confession, the werewolf stayed a werewolf.

The searches against literal shapeshifting returned the Canon Episcopi and Boguet’s treatise. The searches against natural causes returned the medical-history critique and National Geographic’s account of how thin the Stumpp evidence is.

Both searches against opportunism came back empty: one for property confiscation and court fees, one for bite wounds documented before any accusation. The money motive is untested here, not refuted. Three passages were refused by the verbatim gate and count for nothing.

Where the peasants had no Satan, the werewolf stayed a werewolf.

What the number means

Fourteen out of a hundred, and why the number is that low.

Sixteen findings asserted, one conjecture, six open questions. Three findings reached low confidence; the rest computed to unknown.

Falsification scored 18 of 20, the best in the series, because six of eight counter-searches returned something. Corroboration and provenance were close to zero: the whole ledger traced to a single independent origin. The sources are mostly Wikipedia, the Smithsonian and National Geographic, with one systematic review, one peer-reviewed medical history and one primary demonological text.

Saturation scored zero because the second round found three new source families out of five, so the search had not run dry. Nine stress probes were recorded and counter-evidence was found, which caps the result in the unknown band: the components summed to 28, and the cap brings the score to 14.

The counter-searches worked. The sources behind them may still be one story retold.