Lost at Sea: Famous Disappearances Tested against Shipping Traffic

The Bermuda Triangle is said to claim more ships and aircraft than its traffic can explain. I test that claim, and the same claim for every other sea, with open data. From Wikipedia's lists of missing ships, missing aircraft and people who disappeared mysteriously at sea, and the articles on the Bermuda, Dragon's and Michigan Triangles, I built a catalogue of 673 famous disappearances and placed 472 of them at sea from their location text. I compared 27 sea areas, and 2 degree squares, with modern shipping (the World Bank and IMF's AIS ship density, 2015 to 2021) and with 17,224 recorded wrecks that have a position. The Bermuda Triangle holds 8 placed disappearances against 2.8 expected from its traffic (2.9 times; 95% interval 1.2 to 5.6), but its four neighbouring seas have the same excess (2.7 times), and within that group the Triangle holds exactly its share (8 against 7.7). Only 2 of the 14 incidents told in its Wikipedia article lie inside the triangle as usually drawn. The Dragon's Triangle is as expected and the Michigan Triangle holds none. The seas that really are worse than their traffic are Bass Strait and the Tasman Sea, the Great Lakes, the English Channel, the US Atlantic coast, the Caribbean and the Grand Banks, and each has a plain explanation: weather, shoals, historical traffic, aviation and who wrote the losses down. Interactive version: https://maps.tristanhotham.com/lost-at-sea/. Data and code: https://github.com/Tristan-Hotham/lost-at-sea. Blog post: https://blog.tristanhotham.com/2026/10/31/lost-at-sea/. Data: Wikipedia contributors, lists of missing ships, missing aircraft and people who disappeared mysteriously at sea, the Bermuda Triangle, Devil's Sea and Lake Michigan Triangle articles, and lists of shipwrecks (CC BY-SA 4.0; derived tables shared under the same licence); Wikidata (CC0); World Bank, Global Shipping Traffic Density (IMF and World Bank, AIS data 2015 to 2021, CC BY 4.0); Natural Earth.