What was probably the first collection of human skulls for purposes of study was established by Johann Friedrich Blumenbach in Göttingen at the end of the 18th century. In subsequent years, the number of such collections increased, but their importance for scientific research remained modest. A breakthrough took place only in the 1850s when studies on the so‐called cranial index by Karl Ernst von Baer and Anders Retzius gave skull collections a new lease on life, raising physical anthropology from a solely descriptive science to an empirically based “biological anthropology” which used quantitative methods. As a result, Baer was among the first to distinguish between linguistic and morphologic criteria when systematising human populations. This article discusses the development of Baer's ideas concerning the aims and methods of physical anthropology during his career. It pays special attention to the role of scientific collections in moulding Baer's theories.