Reconstructing ship route from amphora origin mapping

What this research found

Where had a fourth-century BCE Greek merchantman wrecked off the Ionian coast of Calabria loaded her cargo, and where was she bound? Amphora typology set a prior over five candidate loading ports, a physics-based square-rig sailing model scored how feasible each approach actually was, and the two were combined into a Bayesian estimate. The most probable answer is a western trader working out of the Corinth-Corcyra corridor in high summer, carrying 14.93 tonnes of wine and oil toward the Bay of Naples.

• The reconstructed cargo is 320 amphorae across seven regional traditions: 230 jars of wine (about 72%) and 90 of olive oil, holding 8,073 litres and weighing 14.93 tonnes with the fired-clay jars included. That fills only about half the 25-30 tonne burden of a comparable small merchantman such as the Kyrenia ship.
• The three leading western candidates for the principal loading port come out statistically close: Corinth and Corcyra at a posterior probability of 0.417, Locri at 0.267, and Syracuse at 0.253. Carthage trails at 0.058 and a north Aegean origin is all but excluded at 0.005.
• Oil punches above its volume. The 80 Corinthian B oil jars lead the manifest by value at 806 drachmas, ahead of the 140 Campanian wine jars at 567, giving a total cargo value near 1,827 drachmas or roughly 0.30 of a talent.
• Summer was fastest and safest on every simulated route, reproducing the ancient preference for high-summer sailing. Nearby Italian and Sicilian ports reach the wreck site in 3.9 to 4.1 days at 8-15% accumulated storm risk, while the long north Aegean haul takes 15.1 days at 34% risk in summer and 17.4 days at 97% in autumn.
• Campania is the most plausible intended market, scoring 0.320 on a composite of wind alignment, transit efficiency, and demand. The onward passage to Neapolis runs 6.1 days at 16% storm risk, against 12.0 and 12.5 days for Rome and Etruria and a 35.2-day beat to windward at 53% risk for Massalia.

How it was done

A representative cargo of 320 amphorae was assembled from the typology and petrography literature, split across seven regional traditions, and given per-jar capacities of roughly 18-32 litres and Attic price bands. A square-rigged holkas was modelled as a performance polar giving through-water speed at each wind angle and strength, with a hard no-go zone inside about 65 degrees off the true wind, a hull-speed cap near five knots, and leeway. That hull was sailed across eight maritime zones carrying seasonal wind and current fields built from modern Mediterranean climatology and adjusted toward early Subatlantic conditions, including a summer Etesian regime scaled up about 12% and a tidal current of nearly three knots through the Strait of Messina. Fifteen voyages covering five origins in three seasons produced transit times and storm risks that were multiplied with cargo share to give posterior origin probabilities, and the simulation was extended past the wreck to four western markets. The output is a 14-page illustrated report labelling every claim as evidence or inference.

Data sources

• Peacock and Williams, Amphorae and the Roman Economy (1986) - amphora typology and fabric framework
• Whitbread, Greek Transport Amphorae: A Petrological and Archaeological Study (1995)
• Koehler, Corinthian A and B Transport Amphoras (1979)
• Casson, Ships and Seamanship in the Ancient World (1995) and Speed under sail of ancient ships (1951)
• Loomis, Wages, Welfare Costs and Inflation in Classical Athens (1998) - Attic price evidence
• Steffy, The Kyrenia ship hull construction, American Journal of Archaeology 89:71 (1985)
• Tammuz, Mare clausum? Sailing seasons in the Mediterranean in early antiquity (2005)
• ORBIS: The Stanford Geospatial Network Model of the Roman World (Scheidel and Meeks, 2012)