Constructing diagrams to understand phenomena and mechanisms

Biologists often hypothesize mechanisms to explai phenomena. Our interest is how their understanding of the phenomena and mechanisms develops as they construct diagrams to communicate their claims. We present two case studies in which scientists integrate various data to create a single diagram to communicate their major conclusions in a research publication. In both cases, the history of revisions suggests that scientists' initial drafts encode biases and oversights that are only gradually overcome through prolonged, reflective re-design. To account for this, we suggest that scientists only develop a unitary understanding of their results through their attempts to communicate them.