Chemical physics, therefore, brings out clearly, or comparatively clearly, something which the ordinary and coarser simplicity-loving theory is obliged and is able to neglect: it realises the virtue that lies in juxtaposition, and shows that the mere outer change of quantity goes with a deeper inward and qualitative one. The result does more than sum up and condense what was spread out in extension and dispersed in parts before: it brings out or reveals something which previously was unsurmised. Always, in a liberal interpretation of the maxim, it is true that _Ex nihilo nihil fit:_ but here, especially, the effect actually discloses what was--but was latent or unperceived--in the premises. The maxim, to be fairly treated, must be read backwards as well as forwards. But we must go a step further if we wish the full explanation. If the premises are to be adequate to support the conclusion, they must be restated in terms which hint at the conclusion--which in a way contain it, but contain it in potentiality and promise, not in act. This is the method of development, which is the method that is applicable to full concrete reality, not like the others to parts abstracted from or insulated in reality. So long as you deal with these selected bits of fact--abstracted from their surroundings, subject to strict observation or strict experiment, you can apply a comparatively simple and straightforward method. You are dealing with abstracted, mutilated, prepared fact. You are guided in these cases by the canons of identity and difference: you add and subtract, or subtract and add; and that is all. You use what are called the rules of experimental method. But these canons do not directly apply--except by happy accident--to the real world, where antecedent and consequents are not separate and tabulated, as the logical canons, the rules of formal logic, require. In dealing with this concrete reality, a much more complex method is needed, a method which has to blend induction with deduction, and to start from both ends in the series of causation at once. You can apply observation or experiment, only when the issues have already been extremely simplified and narrowed down: when the question has been rendered so definite that it is next-door to the answer, and the removal of a slight partition-wall will as it were make the two one clear space. Where observation and experiment are available, indeed, is where the general outlines and principles of the subject are settled, where the scheme of reality is defined in large, but a variety of minor issues still remains to be settled. Unless this general framework is fixed, neither observation nor experiment, with their canon of identity and difference, are of any avail. These methods, therefore, only apply in sciences which are in principle or substantially complete, though admitting of possibly infinite extension in details and particulars. Where the science is yet to constitute, i. e. in dealing with the