Diagrams are ubiquitous today, and we learn already at school how to read and make them. Their epistemic status is, however, curiously ill-defined. This was already the case when the term was first introduced from ancient Greek into Renaissance Latin and into the vernacular languages. In the period 1550-1650, the term “diagram” underwent a complex development, as it came to denote three visually similar types of graphic representations which possessed however quite different degrees of epistemic certainty. The term “diagram” referred, first of all, to mathematical constructions with ruler and compass, like the line constructions that in Euclidean geometry accompany mathematical proofs. Diagram of that first type literally possess demonstrative power, as the concluding words QED (“quod erat demonstrandum”) at the end of a geometrical proof indicate. But quickly, the term “diagram” was also applied to schematic representations, for example in the domain of architectural drawings or machine albums. Diagrams of this type don't yield any proof, but at least provide isomorphically accurate information. Thirdly, "diagrams" also referred to geometrical drawings that represented spatial alterations over time. This last type was often applied to innovative ends, but was clearly more problematic, as it neither furnished any demonstrative proof nor represented existing spatial relations.
I will trace the Renaissance development of “diagram” and show which new possibilities of scientific argumentation it offered, but also, to which controversies its multiple status led. My examples of controversies will be taken from the domain of astronomy (Kepler vs. Fludd) as well as magnetism.
