A second theory holds that metaphor is a useful concession to our feeble imagination. We are all children of Adam to the extent that we crave material embodiments. Even the most highly trained of us are wearied by long continuance of abstract communication; we want the thing brought down to earth so that we can see it. For the same reason that principles have to be put into fables for children, the abstract conceptions of modern science require figures for their popular expression. Thus the universe of Einstein is represented as “like” the surface of an orange; or the theory of entropy is illustrated by the figure of a desert on which Arabs are riding their camels hither and thither. From the standpoint of rhetoric, this theory has some validity. Visualization is an aid to seeing relationships, and there are rhetorical situations which demand some kind of picturization. Many skilled expositors will follow an abstract proposition with some easy figure which lets us down to earth or enables us to get a bearing. There is some value, then, in the “incarnation” of concepts. On this ground alone one could defend the use of metaphors in communication.[167] There is yet another theory, now receiving serious attention, that metaphor is itself a means of discovery. Of course, metaphor is intended here in the broadest sense, requiring only some form of parallelism.[168] But when its essential nature is understood, it is hard to resist the thought that metaphor is one of the most important heuristic devices, leading us from a known to an unknown, but subsequently verifiable, fact of principle. Thus George de Santillana, writing on “Aspects of Scientific Rationalism in the Nineteenth Century,” can declare, “There is never a ‘strict induction’ but contains a considerable amount of deduction, starting from points chosen analogically.”[169] In other words, analogy formulates and to some extent directs the inquiry. Any investigation must start from certain minimal likenesses, and that may conceal the truth that some analogy lies at the heart of all assertion. Even Bertrand Russell is compelled to accept analogy as one of the postulates required to validate the scientific method because it provides the antecedent probability necessary to justify an induction.[170]