PARES Dataset v2 PARES (PArish REcord Survey) contains 535 images of handwritten census tables for years ranging from around 1650 A.D. until 1850 A.D..They come from two French cities, Vic-sur-Seille (French department of Moselle) and Echevronne (French department of Côte d'Or). While they mention very ancient times, the documents are handwritten transcriptions of even older documents and are quite recent, copied from original documents during the 1950's and 1960's for demographic studies led by the INED in France (Institut National des études démographiques − National Institute for Demographic Studies). These copies were made by only a few different writers. The documents are damaged and exhibit different types of degradations. We identified seven different document categories we call C1 to C7. C1 and C3 are generally high-quality documents, without serious damage, consisting of about 90% of the dataset. Other categories include highly damaged documents or documents with specificities. A notable aspect of this dataset is that the records are written using only two different physical paper templates. Categories n°1, 2, 3, 6 and 7 have 25 recordings while the categories 4 and 5 are higher and can record up to 35 recordings. C4 and C5 are the larger templates and differ from the rest of the documents. We published a paper, Text Line Detection in Historical Index Tables: Evaluations on a New French PArish REcord Survey Dataset (PARES), in which we better describe the dataset and the tasks it's possible to run on it.