Methodology

Choosing cases

Cases are famous ones, chosen for how often people look them up and for how much good public-domain material exists about them. The rule of thumb is that the events ended at least 75 years ago and that nobody involved could be alive. There are 95 cases from 1170 to 1949.

Life dates and the "only the dead" check

Every named person in every case (717 people) is checked before the case is published. 448 have a Wikidata item; we read their date of death (property P570) from Wikidata and compare it with the case file. 448 people pass on a Wikidata death date. The remaining 269 have no Wikidata person item (most are victims, witnesses and officials too minor for one) and pass on the record itself, in one of four ways: a death date in the case's sources (for a victim, usually the date of the crime); a birth at least 120 years ago; a case that ended at least 125 years ago; or an adult role at least 100 years ago, which would make them 116 or older today. People who could not be shown to have died by any of these tests are not named; the text describes them by their role instead. A mismatch between the case file and Wikidata, or a person who passes neither test, blocks the case from being published until it is resolved. Wikidata was last read on 26 October 2026.

Contemporary newspapers

Clippings come from Chronicling America, the Library of Congress's archive of digitised American newspapers, through its public loc.gov JSON interface. We use only pages published before 1929, which are in the public domain in the United States. For each case we search short windows around key dates (the crime, an arrest, a verdict), confirm that the page's text really contains the search terms, prefer different newspapers for variety, and then find the column on the page where the terms cluster. That column, with its headline, is cropped from the library's page image; the thumbnail beside it shows where on the page it sits. The excerpt under each clipping is the library's machine-read text, unedited except that it stops before any graphic description. Every set of clippings is looked at before it is published, and off-topic or graphic ones are removed. 58 clippings appear across 12 cases so far; the Library of Congress limits how fast its archive can be searched, so coverage of the other pre-1929 cases is added a little at a time. Requests are made one at a time with pauses and an identifying user agent.

Images

Pictures are period portraits, engravings, buildings and documents from Wikimedia Commons, chosen from each case's and person's Wikidata image and from files picked for the case. A file is used only if its Commons licence allows commercial reuse (public domain, CC0, CC BY or CC BY-SA); each picture is credited with its author, licence and a link to its Commons page. Files whose title, description or categories mention bodies, mortuaries, autopsies, crime scenes or executions are rejected automatically, and every remaining image has been looked at before publication. Images are re-encoded on our own server; no image is loaded from another site. 484 images are in use, under these licences:

• Public domain: 417
• CC BY-SA 4.0: 21
• CC BY 4.0: 10
• CC BY-SA 3.0: 9
• CC0: 7
• CC BY 2.0: 6
• CC BY-SA 2.0: 3
• CC BY 3.0: 3
• CC BY-SA 3.0 de: 3
• No restrictions: 2
• CC BY-SA 2.5: 1
• Attribution: 1
• CC BY 2.5: 1

Photographs are shown with a light sepia tone for the look of the site; hover or tap to see them unfiltered.

Maps

Maps are drawn by our own code from Natural Earth vector data (public domain): coastlines, borders, rivers and modern towns for orientation. The case maps use a locally scaled equirectangular projection; the overview map uses the Equal Earth projection. Coordinates for each place were taken from Wikipedia or Wikidata and checked by hand. No map tiles are loaded from other servers.

Validation: a sample checked against two sources

For 10 cases we checked the key date and the outcome on our page against Wikidata and against Wikipedia or the newspapers of the time. 10 of 10 agree on every check (checked 2026-10-26). Nicola Sacco was first drawn for the sample, but his Wikipedia title redirects to the joint article and has no person item linked to it, so Thomas Neill Cream replaced him.

Sources and licences

• Case accounts: original writing by Sinister Crimes, based on the sources listed on each case page.
• Newspapers: Library of Congress, Chronicling America: Historic American Newspapers; pages before 1929, public domain.
• Life dates and identifiers: Wikidata, CC0.
• Images: Wikimedia Commons; per-image licence (public domain, CC0, CC BY, CC BY-SA), credited on each page.
• Map data: Natural Earth, public domain.

The site is rebuilt every day, which moves "on this day" and the case of the day forward.