L'État et la difficile saisie statistique de la religion : l'exemple des protestants dans les recensements en France au XIXe siècle

As illustrated by the example of the French Protestants, the religious statistics produced by the four censuses held in France from 1851 to 1872 must be taken with great caution. Curiously, in 1851, the distribution of religious denominations by département was not published, due to unspecified «considerations of a particular nature». The question on religion was then removed from the 1856 census before reappearing in 1861, the census administrators noting with satisfaction that the ?quite severe? difficulties encountered in 1851 had by that time disappeared. The stability of the Reformed population count as published, the differences between the enumeration of the urban Protestant population in 1851 and the figures given by local sources, its rapid growth over the subsequent period and the fact that the Reformed population numbers published for this census correspond precisely to the minimum required to justify the number of pastors remunerated by the state under the Concordat suggest that Protestant numbers may have been substantially under-reported. The state may have adjusted the number obtained in 1851 to stave off the threat of politico-religious problems. Under-reporting then decreased from 1861, with a delay of several years in Paris.