Postulate a Design, say “Here was something in the making,” and the process is explicable, especially if fairly rapid so as to bridge over the dangerously weak stages of imperfection. Postulate Natural Selection, and it is manifestly impossible. Now Natural Selection wants that to happen not only with every kind of bird, but with every kind of living creature. (4) _The fourth_ a priori _argument against Natural Selection_: When two or more progenital agents are required, _Natural Selection, acting by blind chance alone, loses effect in geometrical proportion with each generation._ This argument is rather more difficult to follow than the first two, but it is worth understanding, because it is particularly strong, and because it was among the first rude blows against the Darwinian theory. Nägeli brought it out with crushing force as long ago as 1884—it is a commonplace with everyone—except Mr. Wells, who imagines (a great compliment!) that I made it up. Where two or more progenitors are necessary, rare accidental advantages rapidly disappear in a few generations _if the process be left to chance_, as Natural Selection demands. Suppose two progenitors required—as is the case with all animals—there are, of course, many cases in which the total number of factors necessary for the production of progeny is more than two and the argument far stronger, e.g. the pollen of one flower, the pistil of another flower, and the insect which acts as go-between. Take any proportion you like of slightly favoured specimens. Suppose out of a hundred individual males ten show in varying degrees the slight differentiation which gives them a survival-value under changing conditions of environment. It will not be anything like ten out of a hundred, and we have already seen that a _single_ advantage is useless. But we can afford to give this nonsense every advantage in argument, so we will consider only _one_ clear advantage and allow one-tenth of the males to have it. Now, suppose a similar number of females showing in varying degrees this slight valuable differentiation. _Upon the mechanical theory of Natural Selection_, the chances in favour of progeny inheriting that differentiation in the next generation are not one-tenth, but only one-tenth of one-tenth, i.e. one-hundredth. The chances of favoured progeny in the third generation are not one-hundredth, but one in ten thousand. In the fourth, the chances are already only one in a hundred million—which we may call zero.