Tracking patterns and creating clusters: A cross-disciplinary exploration of an extensive collection of materials and historical data from paintings by the Danish artist Vilhelm Hammershøi

The Vilhelm Hammershøi Digital Archive (ViHDA) project, a 5-year research study launched in 2020, is being carried out at the National Gallery of Denmark (SMK) by a cross-disciplinary team of scientists, paintings conservators, imaging specialists, art historians and digital experts. In collaboration with Danish museums and private collectors, as well as museums and owners abroad, who hold works of art by the renowned Danish painter Vilhelm Hammershøi (1864-1916), the project has so far investigated more than 140 paintings by the artist through a systematic visual, technical and scientific examination.
In spite of his growing international reputation, very little was known prior to the present project about the techniques and materials used by Hammershøi. This research aims to map the characteristics and development of his choice of materials and working methods through detailed visual examination and thorough imaging analyses such as infrared reflectography, X-radiography and MA-XRF spectroscopy. To manage the extensive technical and art historical data collected, a relational database was designed to structure metadata, images, text and observations, enabling comprehensive searches across the rich dataset.
This presentation takes a closer look at chemical data from the ground layers in Hammershøi’s paintings, obtained by scanning electron microscopy (SEM) coupled with energy dispersive X-ray spectroscopy (EDXS). Advanced data mining techniques, when applied to the results, uncover patterns and clusters that offer new insights into Hammershøi’s working methods. SEM-EDXS analysis of his preparatory layers has thus identified six categories of ground layers, each with a distinctive pigment composition. When this technical data is combined with art historical research, such as when the paintings were created, where they were painted and what the motif shows, along with thread count data from the X-ray images, questions about Hammershøi’s production, methodological progress and preference for and access to materials can be explored. As an example, a group of paintings known to have been created while Hammershøi resided in Paris in the period 1891 to 1892 is clearly distinct, featuring a unique type of ground and canvas weave. Likewise, weave types in canvases of Hammershøi’s later works have proved to be different from those found in his earlier production. These results not only shed new light on the genesis of Hammershøi’s works, his artistic development and material preconditions, but also serve as a valuable tool when assessing the condition of his paintings.
Upon completion of the project, all data collected during the research will be made publicly available through an open-access digital archive, which will include thousands of high-resolution images available for download and an image viewer designed to allow comparisons of technical images and MA-XRF elemental maps.