10 Life and Letters of Francis Galton
falling vertically through a forest of horizontal pins. He next, starting with a normal distribution of variability a-1, reduces the variability to ru-l by sloping his discharge tubes towards the type (see Fig. 2). This restriction of the tubes has the same effect as giving a uniform horizontal " squeeze " to his original distribution ; he thus reaches his population of " reverted parents." If he now opens any single one of his tubes he will get a normal distribution, about the reverted parent character as type, which will have
Fig. 3. Galton's Quincunx illustrating the effect of Natural Selection.
the proper variability a-1 ./1 - r2 if' a suitable choice be made of the extent of " pin-forest " through which the pellets fall. Since this variability is the same for all parentages, the extent is constant, and if all the tubes be opened, all the " reverted " parentages contribute their share to building up again the population from which we started.
Those who hold the hypothesis of the pure line to be true, apparently overlook the fact that while the gametic distribution might be stable, they must appeal to a stringent natural selection, or a differential fertility, to