In this assumption we postulate for biology the distinctive mathematical concept of infinite divisibility. [29] See Appendix, p. 350. The difference from the mode, or mean, with respect to a definite character in a fully grown organism may be due to the direct action of the environment, in the sense in which we have regarded the environment as influencing the organism; or it may be due to the changes in the organism resulting from the increased or decreased use of some of its parts. The conditions with regard to nutrition, for instance, will not be the same for all the individuals composing a cluster of mussels growing on the sea-bottom. Those in the interior of the cluster do not receive so abundant a supply of sea-water as those on the outside of the cluster; and since the amount of food received by any individual depends on the quantity of water streaming over it in unit time, we shall find that the internally situated individuals will be stunted or dwarfed, while those on the outside will be well grown. Such variations are acquired ones, but even when we allow for them, even if we take care that all the organisms studied live under conditions which are as nearly uniform as possible, there will still be some degree of variability. We cannot be sure that this absolute uniformity ever exists; and the notion of the environment of an organism may be extended so as to include the medium in which embryonic development took place, and even the parental body which formed the environment for the germ-cells from which embryonic development began. But it is probably the case that even with an uniform environment, or with one in which the differences were insignificant, variability would still exist. The variations that might be observed in such a case would belong to two kinds--“fluctuating variations,” and “mutations.” [Illustration: FIG. 21.] Whether the variations observed in a population of organisms are fluctuations or mutations can only be determined by experiment. Let us suppose that we are dealing with a human population, and that the variation studied is that of stature. Let the men with statures considerably over the mean value marry the women who are correspondingly tall, then it will be found that the children from these unions will, when grown up, exhibit a stature which is greater than that of the whole population, but not so great as that of their parents--that is, regression towards the mean of the whole population takes place. This is shown in the above diagram, where the lines above and below the mean one indicate the proportion (relative to the value or frequency of the mean) of people of each grade of stature. The latter is proportional to the distance from the mean measured along the vertical line, distances below this line indicating statures below the mean, and _vice versa_.