Researchers studying science communication often examine how texts addressed to different audiences contribute to the formation of knowledge on a given issue. This article examines how arguments on scientific issues travel from text to text by considering how certain figures of speech persist from version to version. It uses a specialized genre of articles appearing in Science and Nature that introduces research reports appearing later in the issue. These pieces refer explicitly to a research report in the same issue, and in addition to their own agendas, re-present the researchers' claims and supporting evidence. To investigate how the core of an argument survives, the expression claims and lines of support in epitomizing figures are compared. The articles sampled suggest that the figure antithesis, embodying single-difference arguments,often persists from version to version. But in the process of perfecting a figured expression, arguments may be subtly changed in subsequent versions. (Jeanne Fahnestock)