But if the laws of science do not explain what happens--even in the limited sense of scientific explanation--they are the indispensable preliminary to that explanation. If they do not represent the world as it is, they supply the means by which we may hereafter produce the picture. They are ideals not in the sense that science hopes to show eventually that feathers only appear to float leisurely to the ground, and are really all the time falling sixteen feet in a second, but in the sense that starting from the gravitation formula we could show that every feather's fall is as rationally comprehensible as the gravitation formula itself. They are not the ultimate truth, the final reality, or Science's supreme ideal. They are shadows cast by the scientific ideal before its coming; they are the principles by which science must proceed, if she is to make the world of things intelligible. So, too, the reality of the moral ideal does not imply that, in refusing to make sacrifices for others, I only appear to be selfish, and shall be found in the end really to have been actuated all the time by some high moral principle. What is implied is that only by the acceptance of the moral ideal can the world of men be moralised. From the same point of view it seems hopeless to try to make out that atheism, though not in appearance, will be found in reality to have been a manifestation of religion. It is by accepting, not by denying the religious ideal or doubting its existence, that the ideal of religion is achieved. Now, in the theory of evolution we have the attempt made to effect this transition from the abstractions of science to the concrete facts, to show that the world as presented to sense is as intelligible and rationally comprehensible as the laws of science themselves, and that the hypothetical statements of science were but preliminary, though necessary preliminaries, to a categorical statement of actual facts. In evolution, as indeed in all the historical sciences, we abandon the elasticity and the uncertainty of conditional conceptions for the rigidity and certainty of accomplished fact. We no longer deal with what may happen if given conditions are realised, but with what has been, and therefore is subject to no "ifs." We start from the certainty of what is, and thus we argue back positively to what must have been.