Decoding Ancient Economies with RTI and AI

This article explores how numismatists and epigraphers use Reflectance Transformation Imaging (RTI) and coin-die-link matching artificial intelligence (AI) to reconstruct ancient mint production volumes and map historical trade routes. By combining advanced computational photography with machine learning, researchers can capture microscopic surface details on worn artifacts, automate the identification of shared manufacturing dies, and apply statistical models to trace the flow of currency and goods across the ancient world.

Reflectance Transformation Imaging (RTI) is a computational photographic method that captures a subject’s surface shape and color from multiple lighting angles. Researchers place a coin or inscribed artifact in a stationary position and take dozens of photographs while moving a light source to different points inside a virtual hemisphere. Software then combines these images into a single interactive file. This allows epigraphers and numismatists to virtually relight the object from any angle and apply mathematical enhancements to reveal microscopic topography. Through RTI, worn inscriptions, faint mint marks, and subtle die flaws—invisible to the naked eye—become sharp and legible, providing the raw data needed for deeper analysis.

In the ancient world, coins were struck by hand using two engraved metal stamps: an obverse (anvil) die and a reverse (punch) die. Because these dies wore out and cracked from repeated strikes, mints constantly produced new ones. Identifying coins struck from the exact same die—known as a die-link—is essential for understanding mint operations. Historically, finding die-links required researchers to manually compare thousands of coins, a process prone to human error and limited by time. Today, artificial intelligence automates this process. Using computer vision and deep learning algorithms, AI analyzes the high-contrast topographical data generated by RTI scans. The AI measures precise geometric distances between letters, maps microscopic fractures left by damaged dies, and ignores post-mint wear and corrosion. It can cross-reference tens of thousands of coins in minutes to group together specimens that share a common die.

Once the AI establishes the total number of distinct dies present in a surviving sample of coins, numismatists can estimate the original production volume of an ancient mint. Researchers input the AI-generated die-link data into statistical formulas, such as the Esty or Carter methods. These formulas calculate how many original dies likely existed based on the ratio of known coins to known dies in the surviving record. Because metallurgical studies and historical records suggest an average ancient die could strike between 10,000 and 30,000 coins before breaking, multiplying the estimated total number of dies by the average strike capacity yields a reliable calculation of how much money an ancient state injected into its economy during a specific period.

Tracing where these die-linked coins are excavated allows historians to map ancient trade routes with high precision. When epigraphers use RTI to read the mint marks and dates on degraded coins, they establish the exact origin and time of manufacture. When the AI matches a coin found in a distant archaeological site to a die-link chain originating from a specific mint, it creates a physical link between those two geographic points. By plotting thousands of these AI-verified die-links onto Geographic Information Systems (GIS), researchers can visualize the movement of currency. A high concentration of die-linked coins traveling from a central mint to a series of coastal ports reveals maritime trade networks, while chains of chronologically linked dies moving inland often trace the supply lines and campaigning routes of ancient armies.

Together, RTI and AI remove the bottlenecks of manual artifact analysis. RTI extracts the necessary physical data from damaged surfaces, and AI processes that data at scale to find hidden connections, transforming isolated coins into a comprehensive map of ancient economic history.