Abstract
Thought Experiments (TEs) have a long history of use in science. It is suggested that, if science education is to be related as clearly as possible to science, then TEs must play an appropriate part. The relationship between TEs and experiments (Es) is explored. A typology of TEs is presented with examples drawn from the history of physics. The potential uses of the various types of TEs in bringing about conceptual development and as a complement to conventional practical work are addressed. The analysis of three typical school and higher-education level physics textbooks shows that the potentials identified are, at present, not realised. Indeed, elements of TE design were found to be integrated with other pedagogic devices into what we have termed 'thought simulations' (TSs). In these, the behaviour of a phenomenon was illustrated rather than predicted and tested, theory was assumed and embedded rather than being tentative and emergent, and the outcome was assumed rather than being anticipated.