The exhibition “The People of Paris, 1926-1936. In the Mirror of the Census”, recently working till February 8, 2026 on the Musée Carnavalet-Histoire in Paris, is in accordance with the Paris inhabitants censuses of 1926, 1931 and 1936, and contributes to an summary of the capital’s interwar duration.
In France, census operations have been arranged since 1801, but it surely was once now not till 1836 that particular directions got to behave in a uniform means in all of the communes of the rustic. Censuses had been then arranged each 5 years, years finishing at the 1st and sixth, till 1946, except that of 1871, which was once postponed till the next 12 months because of the Franco-Prussian Warfare, and the ones of 1916 and 1941, that have been now not arranged because of the 2 international wars.
Precious knowledge at the inhabitants of Paris
The primary purpose of the inhabitants census is to determine the dimensions of the inhabitants of municipalities for the implementation of a definite collection of regulations. Additionally they permit the number of knowledge on all folks dwelling within the nation at a given time with a purpose to know its construction. Those statistics are compiled from particular person announcements and/or (relying at the 12 months) family sheets (family sheets summarize folks residing in the similar family and their courting with the top of the family) stuffed in via folks and printed in particular publications known as “statistical census results”.
Nominative listing from the 1936 census, inhabitants of ordinary place of dwelling, Saint Gervais district (IV arrondissement). Paris Archives: Reference D2M8 553
Along with those statistics, mayors should bring together a nominative listing of the inhabitants in their municipality. However Paris received the correct to not bring together those lists. Head of the Paris Statistical Place of work, Mr. Lambert, made up our minds to rethink this exception in 1926. The nominative lists of 1926, 1931 and 1936 are subsequently the one ones, with the certainly one of 1946, that exist for the inhabitants of Paris.
If Paris was once given the correct to not bring together those lists, it was once on account of the price of such an operation for this kind of massive inhabitants. Paris’s inhabitants, if truth be told, was once already just about 1.7 million in 1861, one million extra in 1901, and peaked in 1921 at 2.9 million other people. The information contained in those lists are specifically attention-grabbing, as they permit vital refinement of the statistics made within the interwar duration.
A database designed with synthetic intelligence
Those lists, which might be stored within the Paris archives and digitized after which put on-line for roughly ten years, have already researchers who’ve trusted them, as an example, to grasp the evolution of a side road or an area, however have by no means been used as an entire because of the amount that makes their research unimaginable for an remoted researcher. Additionally, short of to paintings from those lists – to start with, to paintings on family construction and particularly on divorced other people – I additionally began counting sure settlements via hand.
Check in of the nominative census listing stored within the Paris Archives and offered within the exhibition “People of Paris, 1926-1936. In the Mirror of the Census”. Carnavalet Museum/Museums of Paris/Pierre Antoine
A gathering with pc scientists from the Laboratory of Computational Sciences, Knowledge Processing and Methods (LITIS), Thierry Paquet, Thomas Constum, Pierrick Tranouez and Nicolas Kempf, consultants in synthetic intelligence, modified the placement as we undertook to create a database of all of the nominal inhabitants lists of Paris from 1936, as a part of the 1936 POP venture. 50,000 photographs, already digitized via the Paris Archives, had been processed the usage of deep finding out and optical persona reputation equipment evolved at LITIS to create the primary database.
The mistakes of this primary “raw” base had been already very small, however we then, with a workforce from the arts and social sciences – made up of Victor Homosexual (Economics College of Toulouse, College of Toulouse Capitole), Marion Letourc (Ined), Joan Davoin (CNRS, Idees), Baptiste Coulmont (ENS Marie Paris-Saclai-Saclai), Pinol (ENS Lyon, Larhra) – nonetheless attempted to right kind mistakes of system studying or column displacement up to imaginable. Those corrections are made robotically, this is, they’re written with a pc script that allows their reproducibility. So, as an example, we modified professions that seem as “benne” to “maid” or “fnene” to “brother”.
Alter the database for statistical research
Then it was once essential to evolve the database to statistical research. Nominative lists, if truth be told, weren’t meant for statistical processing, as a result of they had been shaped without delay from particular person announcements and family information. On the other hand, statistical research calls for that phrases signaling the similar entity be registered in the similar method. This problem is exacerbated in relation to nominative lists: brokers had little house to jot down, since the columns are slender. Due to this fact, they used abbreviations, particularly for longer phrases comparable to beginning divisions or professions.
Due to this fact, we needed to standardize the spelling of all occupations, departments or nations of beginning, family eventualities and names. To do that, we trusted other dictionaries, this is, lists of phrases similar to the variable processed from earlier analysis or different databases. Thus, to right kind the identify of Baptiste Coulmont, who labored in this a part of the database, he used the Insee database for names and deceased individuals. Marion Leturck and Victor Man extensively utilized lists of French departments, colonies and international states, as they had been known as within the interwar duration, or the occupational nomenclature utilized by the Statistics Basic of France (SGF).
In spite of everything, we created lacking variables for the statistical research we would have liked to accomplish, such because the variable “gender” that doesn’t exist within the nominal lists (whilst the ideas seems in particular person recordsdata), and even the delimitation of families with a purpose to perceive their composition. This paintings of correction and adaptation of the database remains to be in development, as paintings is recently being performed on including the nomenclature of occupations – to permit research via skilled teams – or at the introduction of a geographic knowledge machine (GIS) database to succeed in the geolocation of each construction within the town.
In finding ancestors
The POPP base thus created has already been used for more than a few functions. A part of the database (together with surnames – that have now not been corrected -, first names and addresses) was once transferred to the Paris Archives to permit the analysis of nominations in digitized photographs of nomination lists. This new device, which was once put in in the beginning of October 2025 – and in addition introduced for session as a part of the exhibition “People of Paris, 1926-1936 in the mirror of the census” – has already enabled many of us to search out their ancestors.
It was once additionally imaginable for us to give you the first effects taken from the POPP database (in comparison to the statistical result of printed censuses) to create the framework knowledge that looks within the type of infographics within the exhibition (created via Clara Dealberto and Jules Grandin). Those effects additionally seem with a extra comparative viewpoint in a e-newsletter entitled “Paris 100 years ago: a larger population than today and already from elsewhere” (INED, September 2025).
The time has come for the clinical exploitation of the POPP database via the venture workforce, whose goal is to create a portrait of the Parisian inhabitants in accordance with the knowledge to be had within the nominative lists of the inhabitants census, exploring constructions via gender and age, occupation, marital standing, beginning or even the composition of families in several portions of the town.
The writer thank you two different curators of the exhibition “People of Paris, 1926-1936. In the mirror of the census”, Valerie Guillaume, director of the Carnaval Museum – Historical past of Paris, and Ellen Ducate, supervisor of the clinical project on the Carnavalet Museum – Historical past of Paris and the Paris Archives.
