have in like manner been formed by adherence of two or more carpellary leaves, and concludes by pointing out the systematic importance of such considerations. Further on he takes occasion to speak of the significance of the relative number of the parts of the flower, on which head he says much that is good, but does not thoroughly investigate the matter; it was not till a later time that Schimper's doctrine of phyllotaxis made it possible to express these relations of number and position more precisely. He concludes his rules for the application of his morphology to the determination of relations of affinity with the declaration, that the whole art of natural classification consists in discerning the plan of symmetry, and in making abstraction of all the deviations from it which he has described, much in the same way as the mineralogist seeks to discover the fundamental forms of crystals from the many derivative forms. It is obvious that all this teaching was a great step in advance upon the right path, that De Candolle has here given utterance for the first time to an important principle of morphology and systematic botany; nevertheless he did not succeed in always consistently carrying out his own principle; he was true to himself only in the determination of small groups of relationship; in framing the largest divisions of the vegetable kingdom he entirely lost sight of the rule which he had himself laid down, that the morphological character of organs and the extent to which it can be turned to account for systematic purposes is entirely independent of their physiological character, and that the most important physiological characters are just those which are of quite subordinate importance in the determination of affinities. In spite of this strange inconsistency, to De Candolle belongs the merit of being the first to point emphatically to the distinction between morphological and physiological marks, and to bring clearly to light the discordance between morphological affinity and physiological habit; but in this discordance lurks a problem, which could
