Scientific reality is a multi-sided phenomenon which cannot be described in a single and authoritative way. The descriptions of scientific research areas differ if one compares the definitions of science policy programmes with expert judgments in the peer-review process. Bibliometric measurements function as an intermediate representation of science. To make them useful and compatible with other representations they have to be translated. The difficulties of mutual translation of these different delineations of scientific research areas are demonstrated in two case studies (marine sciences and multiple sclerosis research) where each of these three different representations of science is supported by empirical results.