exceedingly difficult thing to distinguish, in the performance of a complex machine, the part played by certain forces, as gravity and inertia, everywhere operating according to known laws. Nevertheless, no one would doubt that gravity and inertia do coöperate in the machine, nor should we for a moment hesitate to ascribe to the one or the other of these two forces whatever subordinate effects are only explainable by each respectively. And so in the present case. Natural selection is not, like the supposititious laws of organic structure, an empirical rule which may to-morrow, perhaps, prove nugatory. Neither, indeed, is it, like mathematico-physical laws, an infallible leading principle governing material events. But, as being a proposition deduced by a chain of valid inferences from universally admitted facts, and at the same time a proposition necessary in se, natural selection stands midway between a rule and a law, though it comes nearer to the latter. Hence, of the two evolution principles of organic nature, laws of structure and natural selection, the latter is in theory the surer, whatever may be its shortcomings in practice.
Undoubtedly it were much to be desired that we could in the individual instance demonstrate the working of natural selection, and follow the process step by step. But this we can not reasonably expect to do. Between the work of natural selection for one generation and the result after 100,000 generations, there subsists about the same ratio as between differential and integral. How seldom it is that we are able to understand this latter ratio, even though we subject it to calculation! But do we for that reason question the correctness of our integration? The corresponding problem, in the present instance, would be to investigate and ascertain the evolution of a species through an endless line of generations and under diverse external conditions, while at the same time, as has been already stated, unintelligible laws of structure, working either not adaptively or only accidentally so, enter the problem as undetermined constants, or even as undetermined functions. Though this can not be done, it does not follow that we must misapprehend the ratio between the differential and the integral found for us by nature, as though by a calculating-machine.
Thus, then, so far as the validity of the principle in general is concerned, it may be for us a matter of indifference whether or not in the individual instance we can discern and demonstrate the operation of natural selection. As things stand, it must be operative, and the only question is, whether its influence, as regards quantity, is comparable to that of the laws of structure, or whether other more powerful influences obliterate its effects, so that the adaptation prevailing throughout nature would be attributable solely to the action of these laws. In view of this question, the following appears to me to be the proper attitude of the investigator of nature:
That natural selection can perform what we must ascribe to it in
