Abstract

One of the distinctive features of George Smith’s work on celestial mechanics is his emphasis on the role of what he calls “second-order phenomena” in the production of high-quality evidence. On Smith’s view, these gaps between theoretical predictions and observations can, under certain circumstances, be a source of evidence far stronger than that achievable through the hypothetico-deductive method. The practice of examining gaps between predictions and observations for the purposes of discovery and testing is commonplace in certain sciences such as seismology, and has played an important role in their development. I use the term reasoning from residuals as a general term for this practice. I think it is worth investigating examples of this set of practices from the history of science, in order to understand the different ways in which reasoning from residuals is done, under what situations it is done, and how it contributes to the growth of knowledge in certain sciences.