This chapter shows how cognitive instrumentalism—the position developed in the previous three chapters—fits with the development of atomic theory between 1885 and 1930, which is typically taken to be a period in which our knowledge of the unobservable ‘realm’ increased considerably. It illustrates how models that were explicitly taken to be non-literal played a significant role in science of the period, and how such models were championed by British scientists of the time as vehicles for understanding. It also shows how this style of modelling led Bohr to his astonishingly fertile model of the atom. It closes by examining the introduction of ‘spin’, and considering, with reference back to Chapter 2, to what extent talk of this should be taken literally.