Painters from medieval times onwards used tricks to improve optical and handling properties of their paints. Historical treatises and manuals on painting techniques provide insights into these methods. Yet, some techniques are rarely mentioned and may represent a mix of standard studio practice, personal preferences, or workshop secrets, and hence were not written down. The extensive presence of ground colourless and sometimes coloured glass particles in paint layers found through scientific analysis belongs to those more mysterious methods. Why did painters add glass to their paints, was it purely economical as it adds bulk, does it impact on the viscosity of the paint, or on transparency? Did they try to imitate glassy effects? We will examine new sampling techniques to extract single particles from paint cross sections, and use analytical methods in a novel application, to establish the chemical composition of the glass, to determine its origin, and indicate possible technical crossovers with the glass and ceramics industries. We will use novel data science on a unique paint cross section database (c. 12,000 samples) from 14th-17th-century European paintings in the Rijksmuseum collection, to build search tools that help us trace and analyse the presence of glass in cross sections in this large dataset. This may tell us where and why artists added glass to their paints and if there is a system to this. Through an interdisciplinary collaboration between data scientists, (technical) art historians, and scientists we want to reveal one of the more elusive painters’ ‘secrets’.