In 1791 France catalogued its stolen libraries on the backs of playing cards
On 15 May 1791 two committees of the French Constituent Assembly — the Comité ecclésiastique and the Comité d’aliénation — printed a fifteen-page pamphlet with a title as long as a paragraph: Instruction pour procéder à la confection du catalogue de chacune des bibliothèques sur lesquelles les directoires ont dû ou doivent incessamment apposer les scellés. An instruction for making the catalogue of each of the libraries on which the district authorities have placed, or must shortly place, the seals.
The seals are the point. France had just nationalised the property of the church, and the libraries of the suppressed religious houses were now state assets sitting in locked rooms in the provinces. Nobody in Paris knew what was in them. The Instruction is a how-to guide, sent to district officials who were not librarians, telling them how to find out.
Its first practical sentence is the famous one:
Avant tout, il faudra qu’ils se procurent une quantité de cartes à jouer suffisante pour y écrire tous les titres des livres.
Before anything else, they must obtain a quantity of playing cards sufficient to write on them all the titles of the books.
This is routinely called the first national union catalogue, and the ancestor of the card catalogue — the drawer of index cards that ran every library on earth until the 1980s. Roughly 1.2 million cards came back to Paris over about three years, describing something over three million volumes. I should say plainly that I could not find an eighteenth-century source for those figures; they are a modern estimate, repeated in one lineage of citations, and I am passing on one attestation rather than two.
The part that is not in the story
Keep reading the Instruction and it stops sounding like a card catalogue.
The cards are not to be used whole. “Ces fichets… se font en coupant une carte dans sa longueur, en deux ou trois parties” — these little slips are made by cutting a card lengthwise into two or three pieces. The popular image, a king of hearts with a book title written on its blank back, is the wrong object. What the officials made were paper strips, and the word for them, fichet, is the ancestor of the French fiche.
Then there is the rule about the author’s name: “On observera scrupuleusement de tirer une ligne sous le nom de l’auteur” — one shall scrupulously observe to draw a line under the author’s name. That underline is a machine-readable field marker, drawn by hand, ninety years before Dewey.
And then this, which is the sentence that changed my mind about the whole episode. Once the slips for a library are written, the cataloguer is told to take a large needle threaded with waxed thread and pierce each card at the bottom — and:
Il faut laisser le fil un peu lâche, pour qu’il y ait du jeu entre les cartes.
The thread must be left a little slack, so that there is play between the cards. (Jeu is both slack and a deck of cards; whoever drafted this had a better ear than the title suggests.)
They sewed them together.
A bound sheaf is a book made of cards
The defining property of a card catalogue is not that it is written on cards. It is that any card can be pulled out and put back somewhere else. That is what makes it a filing system rather than a list: the order is not fixed at the moment of writing, so the catalogue can absorb a new book in 1830 without being rewritten. Cards are the medium; removability is the invention.
The 1791 Instruction takes cards and removes exactly that. Slack thread is not a filing affordance, it is a page-turning affordance — it lets you fan the sheaf to read it, the way you fan a hand of cards. And a further rule sends the numbered slip into the first volume of the book itself, where it works as a shelf marker. The slip is tied to the object, not to a drawer.
So what the districts of France built in 1791 was not the first card catalogue. It was the last hand-written inventory ledger, assembled out of cards because cards were the paper stock to hand, and then bound — because a list you can shuffle was not what anyone wanted from an asset register. You do not want play in an inventory of confiscated property.
Why nobody is arguing about this
I went looking for the scholarly fight over whether the modern card catalogue descends from 1791 and did not find one, which surprised me more than any of the above. Judith Hopkins published “The 1791 French Cataloging Code and the Origins of the Card Catalog” in 1992 and is the obvious anchor for the continuity case; I have not read it and will not summarise a paper from its title. Markus Krajewski’s Paper Machines runs the card index’s history through Vienna around 1780 and Göttingen, not Paris. Kenneth Carpenter dates the American card catalogue to Thaddeus Harris proposing a “slip catalogue” at Harvard in 1840, with no French ancestor mentioned at all.
Three accounts, three different origins, and as far as I can see they are not in dialogue. The disagreement is real but nobody is having it; a reader assembles the conflict out of books that do not cite each other. The French project was never assembled into a national catalogue, and the honest position is that the continuity question — did any of those 1.2 million slips ever reach the hands of a nineteenth-century cataloguer — is not answered in the literature I could reach.
What survives with certainty is smaller and better. Not an institution, a word: fichet → fiche. The revolutionaries cut up a deck of playing cards to inventory a church they had just seized, invented a word for the pieces, and the word outlived the catalogue by two centuries.
Sources: the Instruction itself (Paris, Imprimerie nationale, 1791) via Gallica and the Internet Archive — all French quoted above is from that text. Joseph Smalley’s translation is the standard English version. Figures for cards and volumes are modern estimates from the Library of Congress lineage, not period counts, and I have flagged them as such. Where I have named scholars I have not read, I have said so rather than characterise their arguments.