Various, and often contradictory, hypotheses have been raised for more than a century concerning the composition of ancient varnishes of musical instruments, in particular those made by Antonio Stradivari. Neither these hypotheses, nor the rare experimental results, allow grasping correctly the varnishing processes used by European instrument-makers until the end of the 18th century. The aim of this PhD work is to confront information collected in historical sources with the analysis of a group of instruments varnishes using a specific analytical methodology. First, the study of written and iconographical documents shows that a general and coherent understanding of this topic is not achievable on the basis of these sole elements. Then, a methodology organizing and optimizing chemical analytical methods is elaborated to characterize ancient varnishes of musical instruments. It prioritizes in situ and non destructive analyses, and, when micro-sampling is possible, its analysis is optimized. Thus, techniques as spectroscopies (EDXRF, fluorimetry), micro-spectrometries (FTIR/synchrotron, Raman), separative analyses (GC/MS, Py-GC/MS) and imaging techniques (SEM/EDX, OCT) have been applied to varnishes of a corpus of seventy instruments from European collections, mainly the one in Musée de la musique, Paris. Results deal both with the stratigraphy and the chemical –organic and inorganic– composition of each of the identified strata. These results bring novel insights to the history of varnishing techniques and suggest new approaches to the conservation of these varnished musical instruments and of all their values.