Regionalization Analysis
The Shrinking World
How the Roman communication network contracted, decade by decade, as the empire fragmented into successor kingdoms
Every surviving letter from the late Roman world is a fossilized moment of movement. Someone had to carry it, on foot or horseback, along a specific road or across a specific stretch of sea. The letter itself is just parchment and ink, but the journey it represents reveals the connective tissue of an empire: the roads, the routes, the relationships that kept a vast political entity coherent across thousands of kilometers.
Patrick Wyman’s 2016 USC dissertation, “The Fall of the Roman Empire and the Rise of Its Successor States: Letters, Mobility, and the Transition from Late Antiquity to the Early Middle Ages”, made a deceptively simple argument: if you measure the distances these letters traveled and plot them over time, you can watch the Roman world shrink. Long-distance connections that once spanned the Mediterranean basin gave way to regional networks confined within the borders of Visigothic Spain, Frankish Gaul, or Ostrogothic Italy. The fall of Rome was, at its core, a mobility crisis.
Our dataset of over 7,000 letters lets us extend Wyman’s analysis. Where he focused on the Latin West from 450 to 650 AD with 2,895 letters, we can see the broader picture: both East and West, from the 1st to the 9th century. The chart below tracks the average distance between sender and recipient for every inter-city letter with known coordinates, grouped by decade. We exclude same-city correspondence (where sender and recipient share a location) to isolate the long-range connectivity that defined the Roman communication network.
Communication radius over time
Latin West collections only (450-650 AD), matching Wyman's corpus. Same-city letters excluded. Trend: 50-year weighted moving average.
See it on the map
Watch these letters move across the Mediterranean in real time. The animated map timelapse shows individual arcs between senders and recipients, decade by decade.
Open map timelapseVoices from a fragmenting world
“The roads are blocked, the seas are shut, the provinces are cut off from each other by the movements of peoples hostile and unknown.”
“I have sent this letter by the hand of the deacon Candidus, who goes to manage the patrimony of the blessed Peter in Gaul. Through him, if it please you, send me whatever response you wish.”
“We have been separated by so great a distance that the very occasion of writing seems to have perished, since there is no one traveling from your region to ours.”
Methodology
Distances are calculated as great-circle (haversine) distances between the primary locations of the sender and recipient, as recorded in our authors database. This is an approximation: authors sometimes wrote from locations other than their primary residence, and we assign each person a single coordinate. Still, for the overwhelming majority of letters - especially bishops writing from their sees or officials at their posts - this is a reasonable proxy.
Same-city letters (where sender and recipient share coordinates) are excluded, as they represent local correspondence rather than inter-city mobility. Decades with fewer than 10 inter-city letters are also excluded to avoid misleading averages from small samples. The dot size reflects the number of letters in each decade, and sample sizes are labeled. The trend line is a 50-year weighted moving average that accounts for both temporal proximity and sample size, smoothing decade-to-decade noise to reveal the underlying trajectory. Gaps in the timeline reflect genuine holes in the surviving letter corpus, not suppressed data.
The “Wyman Mode” toggle filters to Latin West collections only (roughly corresponding to the Western Roman Empire and its successor states), within the 450-650 AD range that Wyman studied. This is the subset most directly comparable to his dissertation findings.
Citation: Patrick Wyman, “The Fall of the Roman Empire and the Rise of Its Successor States: Letters, Mobility, and the Transition from Late Antiquity to the Early Middle Ages” (PhD diss., University of Southern California, 2016). Our analysis builds on his methodology with a larger dataset covering both the Greek East and Latin West.