Wednesday, August 14, 2013

Mapping the Great Terror in Moscow

Last week I found out about some data the Russian NGO Memorial compiled of the home addresses of people executed by the NKVD during 1937-38 (the Great Terror) in Moscow.  The original project is available here.  They compiled about 5,500 locations with, by my count, about 12,100 people.  It's pretty terrible stuff.  But it is also an amazing source and something that should be mapped.  And actually I poked around a bit (read: I checked a small link at the bottom of the project's page) and found that someone (Dmitriy Skougarevskiybeat me to it.  His map is good but his map was made ages ago (2011!) so it could use a bit of an update.  He was kind enough to publish the data he used in a Google Fusion Table here. [Update 6/2021: I am migrating all my maps from the beloved and discontinued Fusion Tables to the acceptable ArcGis online. Here is the new version. And here is the heat map version.]  (My table is here, by the way.)  Here is the map I came up with:

I improved on Skougarevskiy's map in two ways: First, his database was missing about 700 addresses that didn't have location data and I was able to find data for most of those (all but nine buildings, where ten executed people lived).  The new map is a mostly complete representation of the data. Second, I wanted to include a way to show not only the points but the number of people arrested, so I used a heat map to visualize it.  Unfortunately, the heat map is not great.  If you zoom in too far, the intensity dissipates and the coloring could be better.  These are limitations of using the very convenient Google applications instead of running the data through your own, nonexistent server.  (Info on the Google Maps API is available here.)  

So what is this data?  Memorial's introduction to the project is useful for understanding the period and what it means for scope and limitations of the data (my translation):

"This list is ordered...not by the place of burial but place of residence: from the addresses from which they [the victims] were taken to die.  Many readers will find here not only their own street, but their house and possible even their apartment...

"The list includes just under twelve thousand people [more, actually SB], surely not a full count, not only because we don't have information about several thousand Muscovites who were killed and then rehabilitated [politically].

"The difficulty is also that in many of the investigative files in the archives no address is shown, or else it is so illegible that it is impossible to connect it to the topography of Moscow.  Almost no one from the list are from those regions of Moscow that were incorporated after 1960 (including Perovo, Liublino, Babushkin, Tsaritsyno and so on).  During the years of repression they were separate settlements with entirely different planning and finding their pre-war location today is almost impossible.  Additionally, thousands of people from city suburbs and other regions of the USSR were executed in Moscow.  For these reasons, this list includes just a third of the larger number of people executed in Moscow for political crimes."

Of course, it goes without saying that this group of people was just a fraction of those executed during 1937-38 overall as well.  Since 1991, historians have learned that in those two years around 700,000 people were executed and 1.5 million arrested by the NKVD at the orders of party leaders, Stalin above all.  Many of these victims were not convicted on the trumped up charges that we might associate with political crime or intellectual opposition to the regime (e.g., as in novels like 1984 or Darkness at Noon).  Instead, they were arrested for the political crime of having been exiled during collectivization as a rich peasant, having been arrested for petty crime too many times or having an ethnic background associated with a bordering state like Poles and Germans.

It is tempting to read hot-spots as buildings where the NKVD projected conspiratorial organizations onto ordinary social networks but it is also possible to read those hot-spots as buildings where lots of people lived.  It does seem like there is a greater intensity of executions in the center of Moscow, where high-ranking party members (hit hard during the terror), would have been more likely to reside.

There are one or two interesting moments:  The famous House on the Embankment (Ul. Serafimovicha d. 2) shows up as bright red, even zoomed in, because 242 people who lived there were executed.  Then there is another area with lots of arrests near metro station Mayakovskaya.  And clicking on point map from there along Tverskaya reveals many buildings where multiple executed people lived.  Perhaps someone with a better knowledge of Moscow's geography in the 1930s can figure has something more to say about the pattern of arrests.

If I have time, I may put out some more maps based on this data to look for other patterns.  Memorial posted a lot of information in the database that Skougarevskiy didn't include in his table. I think it would be particularly neat if I could run through the nationality or place of birth data or party membership from the original data Memorial provides.  Given what is now known about the mass operations in 1937-38, it wouldn't surprise me to see Poland and the Soviet borderlands (pre-war) be a hot spot.  Other thoughts about what might be interesting ways to visualize these records?


  1. excellent project. I'm so glad to know about your work.
    Joan Neuberger

  2. Hello! Thank you for your effort in updating the data. You might take the data with a pinch of salt after this thread: (in Russian).

    1. Thanks for the comment, Dmitrii. I agree that you're not going to get an exact placement of the buildings and that post you linked to shows just why. This is one reason that I generated the heat map in addition to the point map. Even if the addresses are a little off on the heat map, you can still get a pretty decent idea of the density of target areas in Moscow.

    2. By the way, here's a separate exercise by people at NextGIS:
      They included building geometry.