To carry out these experiments with any satisfactory result, it will be necessary to avoid the evil effects of confinement and of too close interbreeding. If birds are experimented with, they should be allowed as much liberty as possible, a plot of ground with trees and bushes being enclosed with wire netting overhead so as to form a large open aviary. The species experimented with should be obtained in considerable numbers, and by two separate persons, each making the opposite reciprocal cross, as explained at p. 155. In the second generation these two stocks might be themselves crossed to prevent the evil effects of too close interbreeding. By such experiments, carefully carried out with different groups of animals and plants, we should obtain a body of facts of a character now sadly wanting, and without which it is hopeless to expect to arrive at a complete solution of this difficult problem. There are, however, some other aspects of the question that need to be considered, and some theoretical views which require to be carefully examined, having done which we shall be in a condition to state the general conclusions to which the facts and reasonings at our command seem to point. _Sterility due to changed Conditions and usually correlated with other Characters, especially with Colour._ The evidence already adduced as to the extreme susceptibility of the reproductive system, and the curious irregularity with which infertility or sterility appears in the crosses between some varieties or species while quite absent in those between others, seem to indicate that sterility is a characteristic which has a constant tendency to appear, either by itself or in correlation with other characters. It is known to be especially liable to occur under changed conditions of life; and, as such change is usually the starting-point and cause of the development of new species, we have already found a reason why it should so often appear when species become fully differentiated.