This essay follows an animated visualization built from Osnabrück’s historic foreigners’ registration index to ask, from the perspective of CRC 1604 ‘Produktion von Migration’, whether a bureaucratic register merely records migration — or helps produce it.

On the screen (see animation below), the year 1943 burns in two colors. Blue points drift in from the west, green points from the east, and both settle into a dark map of Osnabrück that slowly fills with light wherever an address gathers many people at once. During the years of World War II and Nazi rule, the points do not represent ‘migrants’ but deportees, forced laborers brought to the city from the territories the ‘Third Reich’ occupied. A decade later, the palette turns. From the late 1950s, the incoming points run red, yellow, and orange: people moving to Osnabrück now hail from Spain, Italy, Portugal, Yugoslavia, or Turkey, from the UK or the Commonwealth. Labor migration under bilateral recruitment schemes and mobility tied to the large British garrison paint the picture.

Yet, no marker on the screen is a person. Each point is a record, an entry that a municipal office once made about someone it classified as a ‘foreigner’, and what we are watching is not migration but its administrative trace set in motion. We can watch it at all because the entire card index has since been reconstructed as machine-readable data, card by card, in an NGHM-research project at Osnabrück University.¹

This is the center of what follows: an animation built from the Ausländermeldekartei of the city of Osnabrück, the foreigners’ registration index that the city administration kept on paper cards from 1930 to 1980, before its registration system was digitized.¹ The animation belongs to a longer line of work at Osnabrück on the datafication of bureaucracy. Its digitization and analysis are the work of the project “Massendatenbasierte Langzeitmodelle migrationsinduziert wachsender Diversität im urbanen Kontext”, directed by Christoph Rass and Andreas Pott at the University of Osnabrück, funded by the Lower Saxony Ministry of Science and Culture and carried out together with the Niedersächsisches Landesarchiv and the city of Osnabrück.

In 2026, I read this data from the conceptual program of the Collaborative Research Centre 1604, Produktion von Migration. My argument moves from the image to the data behind it, from the data to the project that produced them, from there to the theoretical claim that a register does not record migration but helps produce it, and finally to the question that any such visualization must answer: how does one map an administrative apparatus without reprinting its categories as truth? The animation, I will argue, is most useful precisely where it refuses to behave like a conventional migration map.

Movements to and from and within Osnabrück by inhabitants of non-German nationality 1938-1980 as recorded in the municipal “Ausländermeldekartei” [(c) crass].

What moves on the screen

The animation derives from the digitized registration index an event model of residential movement. Every card records a sequence of addresses with dates; each transition between two addresses becomes one documented movement. The model reconstructs 207,325 such movements for 70,294 people: arrivals in Osnabrück, moves within the city, and departures. More than nine in ten of these movements carry exact coordinates. Osnabrück addresses are placed on the official municipal geocodes, transformed from the UTM grid to latitude and longitude; foreign places are deliberately resolved only to the level of the country.

A word on the period, since the data are uneven. The index was kept between 1930 and 1980, and the movement data became continuously reliable only from about 1938. The animation therefore runs from 1938 to 1980, and it states its window on screen at every moment. A further caution concerns the word ‘migration’ itself, because the index flattens what it gathers. The bright inflows of the war years are not migration in any ordinary sense. Between 1939 and 1945 the cards (partially) register the National Socialist forced-labor system already glimpsed above: the conscripted are fixed to registered addresses dominated by the forced-labor camps erected across the city.¹

What follows is again distinct. The post-war growth records the survivors of that system, first as ‘displaced persons’ and, from 1950, as ‘homeless foreigners’ (Heimatlose Ausländer), the administrative aftermath of a violence-induced mobility rather than a fresh departure. Only from the late 1950s does the index document recruited labor migration, the ‘guest workers’ of the bilateral agreements, a euphemism the National Socialist state had already used to dress coercion as ordinary recruitment.² To let coercion, displacement, and recruitment run under a single heading, as one field of moving points must, is already a category effect: they share an apparatus, not a phenomenon.

The visual grammar of the map animation is austere by design. Each datapoint represents a registered person as a small resting dot at their current address; where many people share an address, a company dormitory, say, the points pile into a bright node, so that brightness reads as occupancy. When someone moves, the point flies along a curved arc to the new address. Around the city sits a ring of countries of origin, arranged roughly by compass direction: arrivals fly inward from the ring, departures fly outward, as two separate, oppositely curved arcs. Color encodes the region of origin throughout. No connecting lines are drawn at all. Density and flow emerge only from the superimposition of many points, in the manner of the particle-flow maps that Max Galka and others made familiar, while the ring borrows the idea of directed in- and out-streams from the circular migration diagrams popularized by Guy Abel and Nikola Sander.³

From card to dataset: the Osnabrück registration project

The animation would be unreadable without the project that turned a drawer of cards into a research dataset. The digitized index comprises some 56,613 mostly double-sided cards holding information on around 70,000 people, and it is the first mass card index of its kind made accessible to research as a digitized dataset.⁴ For the operating period, the digitized dataset documents 66,111 arrivals in Osnabrück, 51,421 departures, and 99,264 changes of address across 10,720 distinct addresses.⁵

The extraction itself was done in 2020 with the AI-service provider omni:us, whose systems read strongly structured documents.⁶ Computer vision built on convolutional neural networks recognized document structure, and here the index posed a particular obstacle: twenty-five different form versions had to be identified before any field could be read. Natural-language processing characterized documents by word markers; recurrent neural networks read the handwritten entries. The raw tables were then rebuilt into a relational database that allows the growth of the index to be simulated over time and individual cards to be reconstructed as digital images, and the Osnabrück addresses were geocoded in cooperation with the city’s geodata service. Automated extraction proved not only faster and cheaper than manual capture but, against expectation, less error-prone, with every data point tied back to the card image for correction.

The pioneer project that led to this outcome is our work on the Osnabrück Gestapo index, which Sebastian Bondzio and I analyzed in a DFG-funded project as a ‘mass data store and world model’ (Massendatenspeicher und Weltmodell): the first attempt to extract by machine the entire data content of an analogue mass index, and an argument that such a card index is less a source about individuals than a body of historical mass data.⁷ The card index, on this reading, was a datafication technology long before the digital era. Data never represent an objective, observer-independent world model, and they never simply speak for themselves.⁸

Our published findings show what the dataset affords. In a 2021 study, we read the index through four cases: the dispersed settlement of arrivals in the mid-1950s, which fits no simple model of ‘arrival quarters’; the sharp post-war growth of those classified as ‘homeless foreigners’ among the displaced; the Latvian ‘watchmen’ whose wartime records were rewritten into employment with the British Army of the Rhine; and more than 3,800 cards for Italians between 1930 and 1980, tracing the fluid passage from ostensibly temporary labor to permanent family settlement.⁹ A separate study followed Spanish migration to Osnabrück between 1960 and 1980.¹⁰ This animation does not replace these close readings, it sets them in motion.

The register does not mirror migration: it produces it

Why insist that the points are records rather than people? Because the conceptual stakes of the CRC 1604 turn on exactly this distinction. Our research program proceeds from the premise that migration is socially produced rather than naturally given, and it borrows Durkheim’s fait social to treat migration, once produced, as a real social fact with effects beyond any actor’s intention.¹¹ It speaks deliberately of production rather than mere construction: it is not only that ideas about migration are socially shaped, but that the reality itself is practically brought about, through apparatuses, procedures, and concrete administrative work. Migration, in this account, is produced through three coupled media: figures (the categories and types through which people are sorted), infrastructures, and spaces.

This is where all three media converge. The Ausländermeldekartei is a data infrastructure, an institutional and material arrangement that enables and constrains action, and at the same time a machine for producing figures, the condensed and socially effective categories through which people are addressed and sorted. It records one figure above all, the ‘Ausländer’, and it records it spatially: the registered are fixed to addresses and tracked through the recorded categories of movement and dwelling, so that the index composes the city itself as a sorted space. Figure, infrastructure, and space meet on a single card. The category ‘Ausländer’ is the clearest case. As the project team noted, it functions only within the logic of the nation-state, by setting the foreigner against the citizen; migration thereby becomes a category of difference, a marker of belonging and non-belonging.¹²

The index does not find foreigners in the world and record them. It is one of the instruments through which ‘the foreigner’ is made administrable in the first place. What we find on the cards are process-generated categories: in Rogers Brubaker’s terms, categories of practice, the working labels administrators use to sort people, rather than categories of analysis, the neutral concepts a scholar would choose. They are bureaucratically produced labels of the kind Roger Zetter described for ‘refugee’, and they operate, as Martina Tazzioli puts it, through a politics of counting.¹³

The history of one word shows how far this reaches. The term ‘guest worker’ (Gastarbeiter) is, as I have argued elsewhere, no innocent word.¹⁴ Max Weber used the term in the 1921 volume Hinduismus und Buddhismus of his Gesammelte Aufsätze zur Religionssoziologie, and may even have introduced it into German scholarly usage, though the attribution is a cautious one; the idea he attached to it was an impossibility of belonging founded on ‘foreignness’.¹⁵ By around 1940 the term had settled into the vocabulary of the National Socialist labor deployment, alongside ‘foreign worker’ and ‘Eastern worker’, as a racialized category of difference. The Osnabrück index registered the people these words were used to sort. To animate the index is therefore to animate a vocabulary, and the points that arrive after 1955 carry a label with a long and compromised past.

The trouble with migration maps

A register of movement invites an obvious visual form: arrows from there to here, flows scaled by volume, a choropleth of origins. That form is a trap. Henk van Houtum and Rodrigo Bueno Lacy have shown how the conventional ‘invasion arrow’ of migration cartography slides from a neutral sign for movement into the iconography of military invasion, composing a threat where it claims only to describe.¹⁶ The straight, smooth, unidirectional arrow assumes a single destination and erases the lived geography and the broken time of actual journeys. The point predates the migration debate. In J. B. Harley’s deconstruction, the map is never a neutral mirror: its rules of objectivity are themselves an exercise of power-knowledge, and the silences of a map are as eloquent as its marks.¹⁷

The most influential modern migration graphic, the circular flow plot that Abel and Sander published in 2014, is elegant and widely imitated, and it is exactly the template one might lay over the Osnabrück data.¹⁸ Yet a chord, a band linking two points on the ring, or an arrow between two places, states a magnitude and a direction while saying nothing true about the path, and it converts an estimate or an administrative tally into a clean, authoritative band. At the same time, the danger for a registration index is specific. Its fields — nationality, occupation, reason for residence — look like hard facts; rendered as flows, they would harden further into a picture of migration as such when they are, in fact, the output of one office’s classifying work.

Designing against the arrow

The animation addresses these objections in its construction rather than in a caption, which is what makes it an argument rather than an illustration. It draws no lines. Movement appears only as points that fly and fade, so that the image never resolves into the false authority of a vector. What looks like density is the superimposition of many resting points, and density measures the frequency of documentation, not historical reality: a bright corridor means a path that was often recorded by an office under a regime of registration. Johanna Drucker’s distinction is the right one here. These are not data, Latin for ‘the given’, but capta, ‘the taken’: constructed, partial, the product of an act of observation, and a humanistic visualization should carry that constructedness rather than smooth it away.¹⁹

Two further choices follow the same logic. The index data available to us ends around 1980, which means that people who left afterward have not had a recorded departure and remain forever ‘present’ in the research dataset: the occupancy field fills up toward the end and never empties. The animation does not correct this artifact, the statistician’s right-censoring, a record cut off before its end, into a tidier curve. It shows the effect and marks it on screen from 1979, because the silence of the source is part of what the source is. And the points stay anonymous. The internal person identifier never leaves the project; only an anonymized movement identifier enters the image, and the animation deliberately declines to reproduce the individual visibility that the registration apparatus was built to achieve. Whose categories are being shown, and who is rendered (in)visible by them, are not incidental questions, as Catherine D’Ignazio and Lauren Klein have argued for data work in general.²⁰ One quarter of the registered persons have no clear region of origin; rather than forcing them into a category, the animation lets them remain dimmed and unresolved.

What becomes visible, and what does not

Within these limits, the moving image makes patterns legible that a table holds only in latent form. Arrivals are dominated by southern Europe throughout the period, with a pronounced peak in 1970, the year of some 4,250 recorded arrivals. The net balance of arrivals over departures runs positive through the boom years and then turns negative in the recession years of 1966–67 and again in 1975, the very years West German recruitment collapsed. Recorded departures swell as the economy contracts: the cyclical steering of labor migration leaves its mark in the registrations themselves. A composition view sharpens the same story into a number. A diversity index, on a scale where 1.0 means many origins evenly mixed and 0 a single dominant group, stands near 0.90 in the early, mixed years, falls to about 0.72 in 1973 as Spanish southern-European dominance concentrates in the inner city, and recovers to about 0.82 by 1980 as Turkish and Yugoslav arrivals re-diversify the picture; in the dense core, orange mixes into red, a handover made visible.

The clearest structural finding concerns how movement inside the city was organized. Intra-urban moves are not diffuse. They are channeled through a few dormitory nodes, the guest-worker residences of the Karmann car factory, and the network analysis identifies the Bohmter Straße 52 as the dominant bridge, a single node through which a disproportionate share of moves passed.

The Weidenstraße 15 (see animation above), the largest of these residences, is the clearest case, and a companion animation follows this single address. In the event model, 4,979 people passed through it, of whom 4,154, more than eight in ten, had their prior registered residence in Spain, above all in Madrid, Valencia, Andalusia, and the small towns of Extremadura; the animation turns this into a visible rhythm, points streaming in from that recruitment geography, gathering at the one address, and then dispersing. It was not a place to live but a place to pass through, an entrance and a junction: of those who moved on, roughly 1,460 stayed in Osnabrück, first in the other Karmann dormitories and then in private flats across the city, while 1,045 returned to Spain and some 1,294 went elsewhere in Germany. It is the guest-worker system in miniature. Its designers intended a rotation, labor that would arrive, work, and leave again, and the building was the architecture of that intention. The dispersal, however, tells another story: the step from the dormitory into the city’s ordinary housing is the quiet beginning of settlement, of the families and futures the rotation principle was meant to prevent, and what the register caught at this address was therefore not transience as such but its conversion into residence, even as the label ‘guest worker’ traveled on with them. The same wave can be watched across the whole city, building and then dissolving, Spanish around 1964, then Turkish in the Südstraße and the Konrad-Adenauer-Ring from 1972–73, a density that migrates through the streets and changes color as it goes.

A second animation sets this case against another, and the comparison is the argument. The Osningstraße, in the Schölerberg quarter, was also a Karmann address, and its dormitory at number 18 housed a comparable, albeit smaller core of Spanish metalworkers passing through; the median stay there, about eleven months, is all but identical to that of the Weidenstraße. By the measure the register makes easiest, duration, the two streets are twins. Yet the Osningstraße (see animation below) is two worlds in one street. Alongside the company dormitory ran a fabric of private tenancies, and through it moved a more mixed population, Spanish but also Portuguese and Yugoslav and, from the early 1970s, increasingly Turkish, with a markedly higher share of women. Here the cards record what the Weidenstraße almost never shows: across the street, some 139 families and 306 children, nearly a quarter of them born in Osnabrück, and a wave of children registered ‘with their parents’ that swells precisely after the recruitment stop of 1973. The same eleven-month rotation, in other words, housed two different futures. At the Weidenstraße it stayed rotation; in the Osningstraße, which will be the subject of one of my coming case studies, it began, quietly, to become settlement, a “second generation” entering the register as foreigners in the city of their birth. The comparison shows what a single figure cannot: that identical administrative durations can hold entirely different social processes, and that the index marks the difference only at its margins, in an entry of ‘Hausfrau’ against an occupation, in a child whose address is given as its parents’.

What the image cannot show is as important as what it can, and its limit is a legal category, not a place. The index holds only those who, at a given moment, lacked German citizenship, and people leave it by changing category as much as by moving away. Naturalization is the clearest case: a registered ‘Spaniard’ who was naturalized became, in the register, a ‘German’ and simply dropped out of the data, without changing address. The gendered citizenship law of the period drew the same line in its own way: until 1953, a foreign woman who married a German man acquired German nationality automatically, and vanished from the index without moving a step. The boundary even ran against birth and descent: before the reform of 1975, a child born in Osnabrück to a German mother and a foreign father was a foreigner too, and so entered the index, not as an absence but as a record. Whether any of this changed how a person was seen on the street, whether the naturalized ‘Spaniard’ stopped being a ‘Spaniard’ to her neighbors, is another question, and one the register cannot answer. Foreign origins, in any case, resolve only to the country, never to the town, and nothing exists after 1980. The animation is, ultimately, a portrait of a registration practice and its reach, complete with the practice’s own blind spots.

The visualization as production

The reflexive consequence is unavoidable, and it should be stated rather than hidden. An analysis of this kind stands inside the process it depicts. To geocode, to assign a region, to color a point by nationality is to extend the classifying work of the office whose cards we are reading, and the scholarship that studies the production of migration also, in a small way, takes part in it. This is why the design choices are not cosmetic: a responsible visualization of an administrative register is one that keeps its own operations visible. The companion argument about large-scale digitization and the silences of the archive, developed for the Paris keynote and published separately on this blog, addresses the same problem from the methodological side.²¹

You may find this animation beautiful, and the beauty is precisely the risk. What looks like a city breathing, alive with diversity, is a city producing it. The light dots are not people arriving. They are a city learning to write down who it would call a foreigner.

---

Notes

1. Sebastian Bondzio, Linda Ennen-Lange, Lukas Hennies, Sebastian Huhn, Max Pochadt, Christoph Rass and Janine Wasmuth, “Die Osnabrücker Ausländermeldekartei 1930–1980. Potenziale als Quelle der Stadt- und Migrationsgeschichte,” Osnabrücker Mitteilungen 126 (2021): 137–193. The paper frames the 1939–1945 entries as the National Socialist forced-labor system and the post-war registrations as displaced persons and “homeless foreigners,” the aftermath of a “gewaltinduzierte[n] Mobilität.” On the latter, see also Lukas Hennies, Sebastian Huhn and Christoph Rass, “Gewaltinduzierte Mobilität und ihre Folgen: ‘Displaced Persons’ in Osnabrück und die Flüchtlingskrise nach dem Zweiten Weltkrieg,” Osnabrücker Mitteilungen 123 (2018): 183–232. ↩
2. Christoph Rass, “Staatsverträge und ›Gastarbeiter‹ im Migrationsregime des ›Dritten Reiches‹. Motive, Intentionen und Kontinuitäten,” in Jochen Oltmer, ed., Nationalsozialistisches Migrationsregime und ›Volksgemeinschaft‹ (Paderborn: Schöningh, 2012), 159–183, DOI 10.30965/9783657773343_010. The Third Reich used bilateral treaties and the euphemism ›Gastarbeiter‹ to cast its wartime deployment of foreign workers as regulated labor migration, even as their treatment converged on that of forced laborers. ↩
3. Max Galka, “Here’s Everyone Who’s Immigrated to the U.S. Since 1820,” Metrocosm (blog), 3 May 2016, metrocosm.com/animated-immigration-map (archived 16 December 2016: Wayback Machine); Will Geary, “TransitFlow,” 2017. On the circular migration diagram: Guy J. Abel and Nikola Sander, “Quantifying Global International Migration Flows,” Science 343, no. 6178 (2014): 1520–1522, DOI 10.1126/science.1248676. ↩
4. Bondzio et al., Ausländermeldekartei (2021), on the number of cards (some 56,613) and persons (around 70,000) and on its standing as the first machine-readable mass card index of its kind. ↩
5. Ibid. Figures from the digitized dataset for the operating period 1930–1980: 66,111 arrivals, 51,421 departures, 99,264 changes of address, 10,720 addresses. ↩
6. Janine Wasmuth and Christoph Rass, “Die Osnabrücker Ausländermeldekartei. Vom Karteigut zum Forschungsdatensatz,” Archiv-Nachrichten Niedersachsen 27 (2023): 42–61. On the extraction with omni:us, the relational database and the geocoding. ↩
7. Sebastian Bondzio and Christoph Rass, “Allmächtig, allwissend und allgegenwärtig? Die Osnabrücker Gestapo-Kartei als Massendatenspeicher und Weltmodell,” Osnabrücker Mitteilungen 124 (2019): 223–260. ↩
8. Sebastian Bondzio and Christoph Rass, “Big Data und die Digitalisierung von Geschichte und Geschichtswissenschaft,” in Stiftung niedersächsische Gedenkstätten, Jahresbericht 2021 (Celle, 2022), 16–21; Christoph Rass and Sebastian Bondzio, “Next Stop Big Data? Erfahrungen mit der Digitalisierung von Geschichte und Geschichtswissenschaft,” in Paul Thomes and Tobias Drewes, eds., Vergangenheit analysieren – Zukunft gestalten, Aachener Studien zur Wirtschafts-, Sozial- und Technologiegeschichte 20, Düren: Shaker, 2020, 13–50). In the original: data “repräsentieren niemals ein objektives, d. h. ein beobachterunabhängiges Weltmodell” (never represent an objective, i.e. observer-independent, world model). ↩
9. Bondzio et al., Ausländermeldekartei (2021), the case studies by Pochadt, Ennen-Lange, Hennies and Wasmuth. ↩
10. Max Pochadt, Christoph Rass and Janine Wasmuth, “Migration aus Spanien nach Osnabrück zwischen 1960 und 1980 im Spiegel der Osnabrücker Ausländermeldekartei: Mobilität, Ankunft und Sozialprofil,” Osnabrücker Mitteilungen 128 (2023): 295–322. ↩
11. Collaborative Research Centre (SFB) 1604 “Produktion von Migration”, University of Osnabrück (funded by the German Research Foundation, DFG). The premises summarized here — production rather than construction, and the triad of figure, infrastructure and space — follow the CRC’s research program. ↩
12. Bondzio et al., Ausländermeldekartei (2021), on the category ‘Ausländer’ as a category of difference within the logic of the nation-state. ↩
13. Rogers Brubaker, Ethnicity without Groups (Cambridge, MA, 2004), on categories of practice versus categories of analysis; Roger Zetter, “Labelling Refugees: Forming and Transforming a Bureaucratic Identity,” Journal of Refugee Studies 4, no. 1 (1991): 39–62, DOI 10.1093/jrs/4.1.39; Martina Tazzioli, “Assembling traces of border violence. Counter-mapping as counter-archiving,” Environment and Planning D: Society and Space (2025), DOI 10.1177/02637758241303367 (on the “politics of counting”). ↩
14. Christoph Rass, “‘Gastarbeiter’. Zur Geschichte eines Schlüsselbegriffs der Produktion von Migration,” NGHM@UOS (blog), 14 June 2020, DOI 10.58079/sbwd. ↩
15. Ibid.; quoted there from Max Weber, Gesammelte Aufsätze zur Religionssoziologie (Hinduismus und Buddhismus), Tübingen 1921, 126 f. Rass states the attribution cautiously (“scheint … eingeführt zu haben,” “vielleicht geprägt”: seems to have introduced it, perhaps coined it). ↩
16. Henk van Houtum and Rodrigo Bueno Lacy, “The migration map trap. On the invasion arrows in the cartography of migration,” Mobilities 15, no. 2 (2020): 196–219 (online 2019), DOI 10.1080/17450101.2019.1676031. ↩
17. J. B. Harley, “Deconstructing the Map,” Cartographica 26, no. 2 (1989): 1–20, DOI 10.3138/E635-7827-1757-9T53. ↩
18. Abel and Sander, “Quantifying Global International Migration Flows” (2014). ↩
19. Johanna Drucker, “Humanities Approaches to Graphical Display,” Digital Humanities Quarterly 5, no. 1 (2011); ead., Graphesis: Visual Forms of Knowledge Production (Cambridge, MA, 2014; metaLABprojects). ↩
20. Catherine D’Ignazio and Lauren F. Klein, Data Feminism (Cambridge, MA: MIT Press, 2020). ↩
21. Cf. the companion post on the DHIP Paris keynote, NGHM@UOS (2026), nghm.hypotheses.org/47456. ↩