One of the art historian's main tasks in heuristics is to gather data in situ, whether of a material, technical, archival or contextual nature. This fieldwork must precede and accompany the establishment & updating of corpuses, inventories, catalogues, etc. This approach, too often overlooked and considered to be part of an outdated and limited art history, is nonetheless essential. It cannot be avoided, at the risk of leading to biased or irrelevant interpretations. The lecture shows the invaluable benefits of this approach, on the basis of examples identified by the two speakers during their research at the Royal Institute of Cultural Heritage (Belgium, Brussels). A systematic approach connected to the object is vital both for fundamental research, by expanding the field of knowledge, and for applied research, involving in particular the (re)discovery of so called masterpieces and the adoption of optimal measures for the conservation and presentation of artworks.