The article reports an empirical study of how authors in popular science magazines attempt to render scientific knowledge intelligible to wide audiences. In bridging the two domains of "popular" and "scientific" knowledge, respectively, metaphor becomes central. We ask the empirical question of what metaphors are used when communicating about modern genetics (DNA and related terms), and what images of human beings are produced. The results show that this field is rich in metaphors and such resources play an important role when recontextualising genetics from one domain to the other. Anthropomorphic metaphors are frequent and genes and DNA are made into intentional agents that decide, choose, and remember. One level of description (molecular) is conflated with a totally different one (intentionality). Other metaphors come from the field of communication (genes as text, letter, library), architecture, and several other fields. The results testify to the role of specific cultural categories when narrating abstract phenomena. It is also argued that popularisation seems to imply making the representation model invisible, and in this sense the model is presented as a claim about ontology. (Contains 10 tables and 3 notes.)