Photo by Bernd Dittrich on Unsplash

Covid-19: the Ports, Roads, Industry hypothesis

The Ports, Roads, Industry hypothesis compiles evidence that there are three discernible characteristics of the spread pattern of Covid-19: port , overland truck routes, and “3rd tier” industrial clusters outside of the big city centers.

Covid-19 is a supply chain disease.

[Check twitter for a friend link to this story]

The hypothesis expressed in a bit more detail:

Covid-19 is concentrated in a mid population density, high small-industry, high particulate matter count, temperate environment.

Ports of entry (seaports, airports, border crossings) are also pronounced patterns.

Most big cities with high rates also follow this pattern. Once we control for population density, few other big cities have the high rates one would expect from high density and high social interaction.

The following collects the factoids in support of the PRI hypothesis, as well as anomalies and contradictions where warranted.

Initial focus is on Italy, Switzerland, Germany, Benelux, and France for now. Additional material for other countries is attached.

Important caveat

We are looking back in time to trace an event chain that happened in January and early February, before there were any known cases of Covid-19 in any of the surveyed countries. This trail is likely no longer active.


The first known case in Italy happened in Casalpusterlengo near Codogno, in the Lodi province of Lombardy — part of the later “red zone”. The poor Google rating is well documented by photography of the interior.

A repeated pattern is local outbreak centers along overland routes: Codogno, Heinsberg, Oise, Igualada, Notably these also contain secondary routes, and apparently only cover a subset of the National route network.

It never really moves from there.

In Northern Italy, the routes from Milan to the Venice, Padova, Treviso, Porto Marghera industrial cluster and from Milan to the port city of Ancona. The continuation from Milan to Turin, home of Fiat, is notably less pronounced. Nothing to Genova. Anecdotally, the hotspots in Veneto are industrial towns, not the old city of Venice.

Dynamic map at observablehq.

The divide between Po plain and green Lake provinces is mirrored in the reported cases by province. C19 is a mid-density and truck route disease.

The pattern still persists, and it never moved beyond the original set of routes. Caveat: case count bubbles don’t account for a variety of things, including density.

C19 is a mid-density Po plain disease in Lombardy: Bergamo, Lodi, Brescia, Cremona. Milan-Monza and the lake provinces have few cases per capita.

In summary, the C19 outbreak was remarkably contained geographically in Italy, even before lockdown conditions. There was enough time to spread beyond the initial areas before the lockdown.

Addendum: timeline of the Lombardy cluster. The initial contagion event likely happened in late January or early February.


Germany has a very particular history with Covid-19, including the contained entry at the Webasto workshop (documented below). The ports and routes pattern is best seen in the picture history of the Wikipedia image page.

Very clear route and industry clusters visible here, especially the A1 from south of the Ruhr area to Hamburg, the A5 from Basel turning right into the A8 near Karlsruhe, the A6 via Nürnberg to Prague, and the string of autobahnen east of Berlin connecting Rostock with Dresden. Also the Württemberg and Bayern clusters.

In this map two industrial clusters stand out: Northeastern Württemberg and Southeastern Bayern. Both are mid-density, highly industrialized areas.

This matches industry and density patterns in Germany, but notably, the big capitals of German automotive world, Munich, Stuttgart, and Wolfsburg are largely unaffected.


There is one obvious port, Saint-Nazaire in the west, plus less pronounced Marseilles in the South, and one apparent industrial cluster, Oise north of Paris. Other than that it’s all routes. Alsace, Calais to Saint-Nazaire circling Paris, the Mediterranean, Rhone, and roughly Clermont-Ferrand to Geneva. Paris is affected but not in relation to its size.


The outbreak centers within the Netherlands follow two secondary truck routes, A73 and more closely N277, past Eindhoven into the suburbs of Nijmegen, and then stop. The port of Rotterdam is mostly unaffected.

A side investigation follows the first detection city of Loon op Zandt to the MICAM she fair in Milan, which also produced he first Portuguese Covid-19 case.


There is a clear division between the industrialized North (Flanders) and the rural South (Wallonia) here, but granularity is not enough to see if spread patterns follow routes. Candidates are all routes connecting the Netherlands with France and the corner near Liege connecting Germany with the Netherlands. Heinsberg is in the continuation of that route.

Additional material

Some additional tweets making peripheral points to the Ports, Roads, Industries hypothesis, and adding evidence to the big city hypothesis (big cities unaffected unless they belong to the PRI clusters). Madrid is an anomaly here.



United States

Of the ten biggest seaports in the United States, five: Seattle, San Francisco/Oakland, Los Angeles/Long Beach, New Orleans, and New York/New Jersey, are outbreak centers. Others like Houston or Savannah seem largely unaffected.

Chicago, Denver, and Atlanta are flight hubs. Detroit is the remaining industrial center. Miami might be affected by cruise ships.


Some evidence that the urban agglomeration Milan-Monza as well as the lake provinces, Como and Varese, were largely unaffected, despite higher population density. Rural areas are also unaffected, Covid-19 is a “mid-density” disease.

Note that three of the four highest density areas of Italy: Naples, Rome, and Genova, are barely affected. C19 follows the mid-density provinces of the North.

Background on Italian textile supply networks.

Particulate matter is a known carrier of Covid-19. The reduction of industrial production could temporarily halt the spread during lockdown.


Some statistical evidence that the 30-million metroplex Tokyo-Yokohama was largely unaffected.



The Webasto cluster was the first detected cluster in Germany, and it was considered contained. This might play a role in the discussion of the S vs L strain hypothesis.

Some circumstantial evidence that Germany doesn’t show high density effects. This should be taken with a grain of salt since it does not consider differences in testing density.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store