The paper discusses some of the problems that may be encountered in writing the history of geology with the help of non-written sources, but also offers suggestions as to the kinds of sources that may prove useful. It considers particularly the well-known proposition of R. G. Collingwood that historical writing should involve the attempted 're-enactment of past experience', and also criticisms of such idealist philosophies of history as have been made by Michel Foucault. In considering the relative merits of these two contrasting views, an example is taken from the author's own work, which illustrates the point that attempted 'thought reading', in the manner commended by Collingwood, can sometimes lead to error; thus Foucault's position gains some support. However, it is contended that so much geological knowledge is specific to particular localities that much of the primary literature may be unintelligible without at least some first-hand knowledge of those localities. Thus, an appropriate mix of written and unwritten sources is necessary for writing the history of geology.