Touch her. If she stops screaming, you did it.

There was no version of this test you could pass.

The touch test played out in front of a packed meetinghouse. An accuser — usually one of the young women at the center of the outbreak — would fall into convulsions on cue, thrashing, crying out that the defendant's specter was attacking her. The court's remedy: bring the accused forward and have her physically touch the afflicted girl.

The folk logic behind it ran through the whole outbreak, not just the courtroom — it's the same belief that drove the earlier ‘witch cake’ incident, where a neighbor baked a cake from the afflicted girls' urine and fed it to a dog, on the theory that harming a piece of the victim would reveal the witch through sympathetic pain. Touch was assumed to be a channel: a witch's power, once expended into a victim, could flow back into her through contact, and the fit would stop.

So it did — reliably. Whether through suggestion, coincidence, or the accuser simply pausing for effect, convulsions in court very often did quiet the instant the accused's hand made contact. To the room, that silence read as confirmation: her power, returning home.

And the alternative was no better. Refusing to perform the test at all was read as a witch too cunning, or too afraid of exposure, to risk the contact. There was no third option — no way to stand in that room and have the outcome read as innocence.

It's worth sitting with what the test actually demonstrates, watched from three centuries away: a legal proceeding that built its central piece of evidence around a performance it could not lose.