The Roman road network created by Itiner-e. Image credits: Itiner-e.

At the height of its dominance, the Roman Empire included over 55 million people, stretching from Britain to Egypt and Syria and covering nearly 4 million square kilometers. In many ways, it was the most impressive Empire in human history. The Romans got many things right, but infrastructure is where they were particularly advanced for the time. In addition to the sea routes, the network of roads across the Empire enabled everything from trading and travel to tax collection. Without it, the Roman Empire couldn’t have lasted.

For centuries, researchers have studied these roads. They’ve mapped them with increasingly better tools. We assumed, with all our satellites and high-tech equipment, that we had a pretty good handle on them. Turns out, we were wrong.

A monumental new study, years in the making, has just presented Itiner-e, the most comprehensive, high-resolution digital map of the entire Roman road network ever created. The findings are staggering. The network is nearly double the length previously cataloged, at 299,171 kilometers (about 186,000 miles) of ancient routes. You can even use the map to calculate how long it would take to get from one city to another on chariot (or donkey).

How Did We Get It So Wrong?

The most detailed resource on Roman roads was the Barrington Atlas of the Greek and Roman World. The atlas was a masterpiece of 20th-century scholarship. But its digital versions, which most modern researchers rely on, were, to be blunt, gross simplifications.

To put it more elegantly, the old maps were of “limited spatial detail”. You had one road going from one city to another in a straight line, ignoring topography and mountains. There’s no way a real road could have passed through there. If you tried to follow it, you’d be driving your chariot off a cliff.

Previously considered roads were often oversimplified, ignoring topography. Image credits: Itiner-e.

Even worse, the sources were opaque. A line on the map wasn’t clearly tied to a specific excavation report or historical text. You just had to trust the cartographer. For scientists trying to build precise models of trade or military movements, this was a disaster.

The creators of Itiner-e, led by Pau de Soto from the Universitat Autònoma de Barcelona, thought this just wasn’t good enough. So, they launched an ambitious, collaborative effort between 2020 and 2024 to synthesize and digitize every known, described, or hypothesized road in the Roman Empire around 150 CE.

Remapping the Roads We Know

The first step was identifying the well-known Roman roads. Even this step was painstaking. They started with things like the Tabula Peutingeriana (a fantastic, bizarre 13th-century copy of a Roman map) and the Antonine Itinerary (a 3rd-century road list) to get the “main highways” and key connections. Next, they used aerial images, starting from historical aerial photographs (like the USAF flights from the 1950s) and moving to modern satellite imagery (Google, ESRI World Imagery), even resorting to declassified Cold War-era satellite imagery (Corona mission) for areas later flooded by modern dams.

It was painstaking work, often requiring the scanning and georeferencing of old maps to align them with modern coordinates.

Examples of the road locating process. The process for creating the Itiner-e dataset involved identifying Roman roads from historical and archaeological sources, locating them using topographic maps and remote sensing, and finally digitizing them using GIS software to create segments with high spatial resolution and detailed metadata. Image credits: Itiner-e.

But even this wasn’t nearly enough. The Romans helpfully left stone markers, or miliaria, all over the empire. The team used a massive geocoded database of 8,388 Latin milestones to get precise, known points on the map. They scanned thousands of regional summaries, excavation reports, and local surveys that don’t get mentioned in large-scale surveys.

They ended up with a mammoth database. But this wasn’t the end of it. Because that’s only the roads we know.

Seeking Out the Ghost Roads

We’ve found a lot of Roman roads. But every year, archaeologists (and sometimes, amateurs) find new ones. There are thousands and thousands of kilometers we haven’t uncovered yet. How do you map those?

Roman roads in busy areas could have directional lanes like here near the ancient city of Timgad in Algeria. Credit: Itiner-e, Artas Media, MINERVA..

To find these ghost roads, the team looked at historical data and overlaid it on modern planimetric data. They looked for faint “linear features” or the ghostly outlines of Roman land divisions (centuriation) still visible in modern farm fields.

But here’s the brilliant part: They also used historical maps. They hunted down 19th and early 20th-century topographic maps because those show the world before modern highways, urban sprawl, and massive dams.

They then digitized every single segment by hand. When a road wasn’t physically visible, they connected the known dots (like milestones or ruins or anything that was linked to a road). But they didn’t just draw a straight line. They followed the most plausible path, hugging the topography. They used digital elevation models to trace the road as it would have actually been built, winding through a mountain pass or following a riverbed.

This step alone can explain why their total road length increased so much. That old “straight line” over a mountain is now, correctly, a realistic, winding 10-kilometer pass.

Illustration of a recently discovered milestone along a mountain road in the rough Cilicia area of Turkey. Credit: Itiner-e.

We Know So Little

The Itiner-e team split the roads into three categories. The first category, the *Certain *ones, were less than 3% of the total roads. These are the roads that have been thoroughly discovered and mapped, and it’s almost nothing.

The bulk of the network were *Conjectured *networks, which made up almost 90%. This means the team had good evidence (from settlements, milestones, or other archaeological findings) that a road was there. So, they were making an extremely educated guess about its exact path. Then, the rest were Hypothetical roads that almost certainly existed between cities, but we haven’t found anything about them yet.

So, why does this matter? Who cares if a 2,000-year-old road was 500 meters to the left or right of where we thought?

Well, for starters, researchers can now model ancient trade with far greater precision. How long really did it take to get wheat from Egypt to Rome, or wine from Gaul? Our previous models were almost certainly imprecise, and this makes it much easier to calculate. It’s not a straight line. It’s a winding path with an average slope (which, by the way, the team also calculated for every single segment). This changes all our calculations about cost, time, and economic integration.

It also changes our understanding of Roman warfare. Now, a military historian can model, almost to the hour, how long it would take a legion to march from a fortress to quell an uprising, adapting to the actual terrain.

Plus, the dataset is open access. Every time archaeologists find something new, they can add it to the database. The team also created “confidence maps” that show which regions are well-studied and which are not — a literal treasure map for future archaeologists, showing exactly where to dig next.

The study was published in Nature.