The genes whose existence has been established in this way have many peculiar properties. They are extremely small bodies ; their centres need be no more than four protein molecules apart, so that they must be composed of so small a number of molecules that the ordinary laws of physics and chemistry, which are statistical statements applicable oniy to bodies containing great numbers of molecules, need not necessarily be accurately followed by them.
The genes are clearly capable of growth and division, because many hundreds or thousands of cell divisions lie between the original zygote nucleus and the gametes which ultimately arise from it. They are individual, and extremely stable because they must persist throughout long evolutionary chains. Nevertheless, they must be capable of such modification as is necessary to cause the appearance of a mutation. It is possible that they owe this power to their minuteness. Thus genes present many of the char acters of living organisms, in at least as high a degree as cells themselves. It has, in fact, been suggested that they are similar to the elements of the viruses which are responsible for many diseases.
It now remains to consider the possibility of a hereditary trans mission of continuous variations. We have already seen that it is not certain that any such exist, and a consideration of the nature of any conceivable hereditary mechanism rather suggests that they do not. It is certain that the whole mechanism which determines the course of development exists in the fertilized egg, in part in the nucleus, in part certainly in the cytoplasm. Any congenital variation must result from some change in this mechanism, and such change must, in the end, be of a "chemical nature," including under that head all modifications of the relationship of the mole cules with each other. But such changes must, in an ultimate
analysis, be definite steps, not connected with one another by in termediates. Thus it is difficult to believe in the existence of true heritable continuous variations.
The only possible mode of investigating the inheritance of an apparently continuous variation is a statistical one. Two indi viduals which differ from one another are crossed and their off spring bred together, either with or without selective mating. The character under investigation is measured in the original par ents and in all their descendants, and by an appropriate technique mathematical expressions of their resemblance to the original parents and to each other can be established. The results so ob tained may be of very great value in discussions of the probable course of change of a population, which is the basis of evolution.
But the mode by which these results are obtained is such that they can tell us nothing about individuals; they are applicable only to population. (See BIOMETRY.) These results are probably not much less valuable for their spe cial purpose if the variation considered be indeed dependent on a large number of genes, each exhibiting a Mendelian inheritance, and not truly continuous. The existence of this condition may be suggested by the fact that, if it obtains, the first hybrid generation will exhibit a smaller range of variation than that of the popula tion of which the original parents were members, and that a long continued course of inbreeding will much reduce the variability of the resulting population.