complex, and it would seem as though the system would soon fall of its own weight. The entire value of the hypothesis consists in the formal approximate expression of certain facts; when it is found that the hypothesis begins to fail even for the classes of facts for which it was originally intended, and that most of the known facts of development can not possibly be expressed in its terms, the entire conception is put on trial.
The weakness in the theory of unit characters is in the use and conception of the term "character.' The term has been prescribed to us by the systematic zoologists and botanists engaged in describing the differences between species; so that "character" really means any definable feature of an anatomical kind that differentiates species; by extension it also means any other differentiating features that can be defined. In the study of evolution and heredity, it is usually only anatomical characters that are in question. Now the study of the physiology of development teaches us that whatever else "characters" may be, they are not units; they simply represent the sum of all physiological processes coming to expression in definable areas or ways, and they may thus represent a particular stage of a chemical process, or a mode of reaction of some part. "Character" is essentially a static morphological term; in the study of heredity and development we are dealing with biological processes. To adapt a phrase of Huxley's: "characters" are like shells cast up on the beach by the ebb and flow of the vital tides; they have a more or less adventitious quality. To give them representation in the germ is equivalent to a denial of uniformity in biological phenomena.
Just as the exact position of each shell on a beach might be fully explained if we knew its full history, so each character has a certain kind of inner necessity as the result of a sequence of developmental processes. And just as in the history of the position of the shell on the beach we should certainly ascribe great importance to the tides and winds, so in the quality of each individual character we should find corresponding vital tides and winds, as regular and lawful as those of the ocean. We do not yet know the secrets of the vital tides; we maintain only that they are the moving forces in development and heredity, just as in physiology and pathology; and every fundamental contribution to the physiology of protoplasm is at the same time, and to the same extent, a contribution to heredity and the physiology of development.
But if these principles are accepted, how are we to explain the facts on which the theory of unit characters depends? The main difficulty lies not in the facts of mutation, for the physiology of this phenomenon already begins to appear from the experiments of Tower and MacDougal, who show that mutations may result from action of
