The existence of differences of form introduces a new problem, the problem of diversity. There are only a few possible combinations of the principal organs, but as you get down to less important parts the possible scope of variation is greatly increased, and most of the possible variations do exist. Nature seems prodigal of form, of form which needs not to be useful in order to exist. "It needs only to be possible, _i.e._, of such a character that it does not, destroy the harmony of the whole."[54] We seize here the relation of the principle of the adaptedness of parts to the problem of the variety of form. The former is in a sense a regulative and conservative principle which lays down limits beyond which variation may not stray. In itself it is not a fountain of change; there must be another cause of change. This thought is of great importance for theories of descent. Cuvier has no theory to account for the variety of form: he contents himself with a classification. There are two main ways of classifying forms; you may classify according to single organs or according to the totality of organs. By the first method you can have as many classifications as you have organs, and the classifications will not necessarily coincide. Thus you can divide animals according to their organs of digestion into two classes, those in which the alimentary canal is a sac with one opening (zoophytes) and those in which the canal has two openings,[55] a curious forestalment, in the rough, of the modern division of Metazoa into Coelentera and Coelomata. It is only by taking single organs that you can arrange animals into long series, and you will have as many series as you take organs. Only in this way can you form any _Echelle des etres_ or graded series; and you can get even this kind of gradation only within each of the big groups formed on a common plan of structure; you can never grade, for example, from Invertebrates to Vertebrates through intermediate forms[56] (which is perfectly true, in spite of Amphioxus and Balanoglossus!). In the _Regne Animal_ Cuvier restricts the application of the idea of the _Echelle_ within even narrower limits, refusing to admit its validity within the bounds of the vertebrate phylum, or even within the vertebrate classes. This seems, however, to refer to a seriation of whole organisms and not of organs, so that the possibility of a seriation of organs within a class is not denied. Cuvier was, above all, a positive spirit, and he looked askance at all speculation which went beyond the facts. "The pretended scale of beings," he wrote, "is only an erroneous application to the totality of creation of partial observations, which have validity only when confined to the sphere within which they were made."[57] This remark, which is after all only just, perfectly expresses Cuvier's attitude to the transcendental theories, and was probably a protest against the sweeping generalisations of his colleague, Etienne Geoffroy St Hilaire.