Cold Years, Witch Years? Scotland's Witch Panics and the Little Ice Age, 1563 to 1736

Scotland's witch-hunt came in bursts. Using the Survey of Scottish Witchcraft, I count 3,840 people accused between 1563 and 1736 (3,413 cases, 3,212 named accused) and map 3,133 of the named accused through residence coordinates held in Wikidata. Five recognised panics (1590 to 1591, 1597, 1628 to 1631, 1649 to 1650 and 1661 to 1662) cover 11 of 174 years but hold 57% of dated accusations. 85% of named accused of known sex were women, a share that did not change between panic and quieter years. Accusations were concentrated in the Lowlands around the Firth of Forth (54% of mapped accused within 50 km of Edinburgh), but each panic had its own heartland. I then test the Little Ice Age hypothesis, that witch-hunting followed cold years, against reconstructed European seasonal temperature (Luterbacher et al. 2004; Xoplaki et al. 2005), using tests that respect the small number of panics and the autocorrelation of both series. None of 30 annual correlations reaches p < 0.05 under circular shifts; the panic onsets were not colder than randomly chosen years; the coldest tenth of years saw fewer accusations than expected; and a weak decadal association (r = −0.21, p = 0.15) reverses sign when the eighteenth century is excluded. The only hint, cool summers before panic onsets, does not survive a family-wise check (p = 0.25). Scotland's panics do not appear to have been set off by Europe-wide cold, though the test cannot exclude local weather effects. Interactive version: https://tristan-hotham.github.io/witch-hunts/. Data and code: https://github.com/Tristan-Hotham/witch-hunts. Blog post: https://blog.tristanhotham.com/2026/10/27/witch-hunts/.