(3) It is sometimes said, in bold metaphor, that natural selection is constantly on the watch to select any modification, however slight, which is of advantage to the species. And it is true that elimination is ceaselessly operative. But it is equally certain that the advantage must be of sufficient value to decide the question whether its possessor should be eliminated or should escape elimination. If it does not reach this value, Natural Selection, watch she never so carefully, can make no use of it. Elimination need not, however, be to the death; exclusion from any share in continuing the species is sufficient. To breed or not to breed, that is the question. Any advantage affecting this essential life-function will at once catch the eye of a vigilant natural selection. But it must be of sufficient magnitude for the machinery of natural selection to deal with. That machinery is the elimination of a certain proportion of the individuals which are born. Which shall be eliminated, and which shall survive, depends entirely on the way in which the individuals themselves come out in life's competitive examination. The manner in which that examination is conducted is often rude and coarse, too rough-and-ready to weigh minute and infinitesimal advantages. What must be the value of a favourable or advantageous modification to decide the question of elimination, to make it an _available advantage_, must remain a matter of conjecture. It will vary with the nature and the pressure of the eliminative process. And perhaps it is scarcely too much to say that, at present, we have not observational grounds on which to base a reliable estimate in a single instance. We must not let our conviction of its truth and justice blind us to the fact that natural selection is a logical inference rather than a matter of direct observation. A hundred are born, and two survive; the ninety-eight are eliminated in the struggle for existence; we may therefore infer that the two escaped elimination in virtue of their possession of certain advantageous characters. There is no flaw in the logic that has thus convinced the world that natural selection is a factor in evolution. But by what percentage of elimination-marks the second of the two successful candidates beats the senior on the list of failures we do not know. We can only see that, on the hypothesis of natural selection, it must have been sufficiently appreciable to determine success or failure.