The paper proposes a sequence of instructions that corresponds to the method of explanation in its ideal form. The method of explanation is not analytic. Nevertheless, its particular executions may be analytic without affecting its specific cognitive goal (the growth in understanding). Therefore, the method is characterized as “potentially analytic”. Drawing on Zeleňák’s critique of a purely causal view of the explanation relation, as well as on some arguments against Zeleňák’s “mixed view”, the paper argues for a view of the explanation relation as obtaining between abstract objects (the explanans and the explanandum). In the classic case, these are propositions: what is described by (the proposition in) the explanans explains what is described by (the proposition in) the explanandum.