The Limits of Concept Formation in Natural Science