This dissertation examines how scientists and scientific editors have approached specific problems related to visualization and visual argumentation in scientific texts. These problems are related to the following research questions: (1) How are new visualization practices established as scientifically credible? (2) How do scientists modify existing instrument output to make new visual arguments? (3) How do scientists use verbal and visual means to transform problematic data into acceptable support for novel claims? (4) What are the practical and ethical boundaries of modifying visual artifacts for scientific arguments? (5) How do scientists refute established (but incorrect) visualizations that have been widely accepted as accurate representations of reality?

This project considers these issues rhetorically by examining a number of recent and historical cases. The first three case studies explore how scientists created both compelling and uncompelling visual arguments by mediating the visual output of instruments with rhetorical strategies. These case studies focus on visualizations from physical science: x-ray diffraction photographs, graphics establishing the theory of plate tectonics, and visualizations of atmospheric phenomena. In each case, visualizations articulated invisible phenomena in new ways, transforming unclear or seemingly unremarkable data into convincing knowledge claims. My analysis of these cases explores how scientists integrate visuals into the analogical, causal, transitive, symmetrical, and dissociation arguments that are so essential to the practice of science. The later case studies examine broader concerns regarding ethics, persuasion, and modern scientific visualization. I examine recent issues related to the digital generation and manipulation of scientific images and rhetorical issues related to scientists' increasing dependence on complicated computer algorithms for creating visual arguments.