So far we have considered Geoffroy's theories in their application to the facts. We go on to discuss the theories themselves, and the general conception of living things which underlies them. The principle of unity of plan and composition is the keynote of Geoffroy's work. It states that the same materials of organisation are to be found in all animals, and that these materials stand always in the same general spatial relations to one another. The "materials of organisation" are not necessarily organs in the physiological sense, and indeed the principle of the unity of plan cannot be upheld if the unity has reference to organs only. This became clear to Geoffroy, especially in his later years. In 1835 he wrote, speaking of the principle of the unity of plan, "I have, moreover, regenerated this principle, and obtained for it universality of application, by showing that it is not always the organs as a whole, but merely the materials composing each organ, that can be reduced to unity."[109] Even in the _Philosophie anatomique_ he deals rather with parts than with organs; he deals, for instance, with the elementary parts of the sternum, not with the organ "sternum" in its totality. The functions of the sternum vary, and the primary protective function of the sternum may be assumed by quite other parts, _e.g._, by the clavicles in fish, which protect the heart.[110] True homologies can be established between materials of organisation but not always between organs, which may be composed of different "materials." Almost as a corollary to this comes the further view that form is of little importance in determining homologies. An organ is essentially an instrument for doing a particular kind of work, and its form is determined by its function. Organs which perform the same function are usually similar in form though the elementary materials composing them may be different. This is seen in many cases of convergence. Organs, therefore, which perform the same function and are similar in external form are not necessary homologous. Conversely, the same complex of materials, say a fore limb, may take on the most varied shapes according as the function of the organ changes--but homology remains though form changes. Accordingly, form is one of the least important elements to be considered in determining a homology. "Nature," he wrote in one of his early papers, "tends to repeat the same organs in the same number and in the same relations, and varies to infinity only their form. In accordance with this principle I shall have to draw my conclusions, in the determining the bones of the fish's skull, not from a consideration of their form, but from a consideration of their connections."[111] Again, after comparing a vertebra of the Aurochs with an abdominal segment of the crab, he says, "I have insisted upon an identity which has extended to the least important relation of all, that of form."[112]