Spencer defined life as the continuous adjustment of internal relations to external relations. It is one of the best that has been given, though I am not sure that Professor Brooks has not improved upon it when he says that life is "response to the order of nature." This seems a long way from the definition of Verworn, heretofore cited, as the "metabolism of proteids." To this Brooks opposes the telling epigram: "The essence of life is not protoplasm but purpose."

Without attempting adequately to illustrate the nature of organic adaptations, I will direct your attention to what seems to me one of their most striking features regarded from the mechanistic position. This is the fact that adaptations so often run counter to direct or obvious mechanical conditions. Nature is crammed with devices to protect and maintain the organism against the stress of the environment. Some of these are given in the obvious structure of the organism, such as the tendrils by means of which the climbing plant sustains itself against the action of gravity or the winds, the protective shell of the snail, the protective colors and shapes of animals, and the like. Any structural feature that is useful because of its construction is a structural adaptation; and when such adaptations are given the mechanist has for the most part a relatively easy task in his interpretation. He has a far more difficult knot to disentangle in the case of the so-called functional adaptations, where the organism modifies its activities (and often also its structure) in response to changed conditions. The nature of these phenomena may be illustrated by a few examples so chosen as to form a progressive series. If a spot on the skin be rubbed for some time the first result is a direct and obviously mechanical one; the skin is worn away. But if the rubbing be continued long enough, and is not too severe, an indirect effect is produced that is precisely the opposite of the initial direct one; the skin is replaced, becomes thicker than before, and a callus is produced that protects the spot from further injury. The healing of a wound involves a similar action. Again, remove one kidney or one lung and the remaining one will in time enlarge to assume, as far as it is able, the functions of both. If the leg of a salamander or a lobster be amputated, the wound not only heals but a new leg is regenerated in place of that which has been lost. If a flatworm be cut in two, the front piece grows out a new tail, the hind piece a new head, and two perfect worms result. Finally, it has been found in certain cases, including animals as highly organized as salamanders, that if the egg be separated into two parts at an early period of development each part develops into a perfect embryo animal of half the usual size, and a pair of twins results. In each of these cases the astonishing fact is that a mechanical injury sets up in the organism a complicated adaptive response in the form of operations which in the end counteract the initial mechanical effect. It is no doubt true that somewhat similar self-adjustments or responses may be said to take place in certain non-living mechanical systems, such as the spinning top or the gyroscope; but those that occur in the living body are of such general occurrence, of such complexity and variety, and of so design-like a quality, that they may fairly be regarded as among the most characteristic of the vital activities. It is precisely this characteristic of many vital phenomena that renders their accurate analysis so difficult and complex a task; and it is largely for this reason that the biological sciences, as a whole, still stand far behind the physical sciences, both in precision and in completeness of analysis.

What is the actual working attitude of naturalists towards the general problem that I have endeavored to outline? It would be a piece of presumption for me to speak for the body of working biologists, and I will therefore speak for only one of them. It is my own conviction that whatever be the difficulties that the mechanistic hypothesis has to face, it has established itself as the most useful working hypothesis that we can at present employ. I do not mean to assert that it is adequate, or even true. I believe only that we should make use of it as a working program, because the history of biological research proves it to have been a more effective and fruitful means of advancing knowledge than the vitalistic hypothesis. We should therefore continue to employ it for this purpose until it is clearly shown to be untenable. Whether we must in the end adopt it will depend on whether it proves the simplest hypothesis in the large sense, the one most in harmony with our knowledge of nature in general. If such is the outcome, we shall be bound by a deeply lying instinct that is almost a law of our intellectual being to accept it, as we have accepted the Copernican system rather than the Ptolemaic. I believe I am right in saying that the attitude I have indicated as a more or less personal one is also that of the body of working biologists, though there are some conspicuous exceptions.

In endeavoring to illustrate how this question actually affects research I will offer two illustrative cases, one of which may indicate the fruitfulness of the mechanistic conception in the analysis of complex and apparently mysterious phenomena, the other the nature of the difficulties that have in recent years led to attempts to re-establish the vitalistic view. The first example is given by the so-called law or principle of Mendel in heredity. The principle revealed by Mendel's wonderful discovery is not shown in all the phenomena of heredity and is probably of more or less limited application. It possesses however a profound significance because it gives almost a demonstration that a definite, and perhaps a relatively simple, mechanism must lie behind the phenomena of heredity in general. Hereditary characters that conform to this law undergo combinations, disassociations and recombinations which in certain way suggest those that take place in chemical reactions; and like the latter they conform to definite quantitative rules that are capable of arithmetical formulation. This analogy must not be pressed too far; for chemical reactions are individually definite and fixed, while those of the hereditary characters involve a fortuitous element of such a nature that the numerical result is not fixed or constant in the individual case but follows the law of probability in the aggregate of individuals. Nevertheless, it is possible, and has already become the custom, to designate the hereditary organization by symbols or formulas that resemble those of the chemist in that they imply the quantitative results of heredity that follow the union of compounds of known composition. Quantitative prediction—not precisely accurate, but in accordance with the law of probability—has thus become possible to the biological experimenter on heredity. I will give one example of such a prediction made by Professor Cuénot in experimenting on the heredity of color in mice (see the following table). The experiment extended through three generations. Of the four grandparents three were pure white albinos, identical in outward appearance, but of different hereditary capacity, while the fourth was a pure black mouse. The first pair of grandparents consisted of an albino of gray ancestry, AG, and one of black ancestry, AB. The second pair consisted of an albino of yellow ancestry, AY, and a black mouse, CB. The result of the first union, AG x AB is to produce again pure white mice of the composition AGAB. The second union, AY x CB is to produce mice that appear pure yellow, and have the formula AYCB. What, now, will be the result of uniting the two forms thus