Abstract
The analysis of texts from the domain of natural science indicates that the metaphor constitutes one of their most significant components and one of the fundamental tools for communicating new information and explaining phenomena. Further research on this issue ought to include a more formalized classification of the functions of metaphor in scientific texts, because the divisions suggested by Paton and Boyd are nor fully separate and the types of metaphor they proposed overlap. It shall also be necessary to reformulate the issue of similarity between carrier and topic in the existing definitions of metaphor, because otherwise the majority of phenomena described in the analyzed texts (except for very conventional expressions such as cerebellum) will not qualify to be called metaphors.