Particle.news

AI Helps Reconstruct Rules of Lost Roman Board Game From Wear Patterns

AI simulations matching wear on a Roman stone board point to a blocking game now recreated for public play.

Overview

  • Archaeologist Walter Crist identified a 21×14.5 cm limestone plate with carved lines in the Roman baths archive at Heerlen, ancient Coriovallum.
  • Microscopy and high‑resolution 3D scans dated the object to roughly 1,500–1,700 years ago and found concentrated wear along the engraved lines.
  • Researchers ran two AI agents trained on later European rule sets to test gameplay patterns against the board’s wear traces.
  • The best fit was a blocking‑style game in which pieces maneuver to immobilize an opponent, pushing evidence for such games back into the Roman period.
  • The team has posted a playable reconstruction called Ludus Coriovalli and says a full report is slated for the journal Antiquity.