This was succeeded by Rev. J. T. Gulick's profound essays "On Diversity of Evolution under One Set of External Conditions"[235] (1872), and on "Divergent Evolution through Cumulative Segregation"[236] (1887). These and later papers are based on his studies on the land shells of the Hawaiian Islands. The cause of their extreme diversity of local species is, he claims, not due to climatic conditions, food, enemies, or to natural selection, but to the action of what he calls the "law of segregation." Fifteen years later Mr. Romanes published his theory of physiological selection, which covered much the same ground. A very strong little book by an ornithologist of wide experience, Charles Dixon,[237] and refreshing to read, since it is packed with facts, is Lamarckian throughout. The chief factor in the formation of local species is, he thinks, isolation; the others are climatic influences (especially the glacial period), use and disuse, and sexual selection as well as chemical agency. Dixon insists on the "vast importance of isolation in the modification of many forms of life, without the assistance of natural selection." Again he says: "Natural selection, as has often been remarked, can only preserve a beneficial variation--it cannot originate it, it is not a cause of variation; on the other hand, the use or disuse of organs is a direct cause of variation, and can furnish natural selection with abundance of material to work upon" (p. 49). The book, like the papers of Allen, Ridgway, Gulick, and others, shows the value of isolation or segregation in special areas as a factor in the origination of varieties and species, the result being the prevention of interbreeding, which would otherwise swamp the incipient varieties. Here might be cited Delboeuf's law:[238] "When a modification is produced in a very small number of individuals, this modification, even were it advantageous, would be destroyed by heredity, as the favored individuals would be obliged to unite with the unmodified individuals. _Il n'en est rien, cependant._ However great may be the number of forms similar to it, and however small may be the number of dissimilar individuals which would give rise to an isolated individual, we can always, while admitting that the different generations are propagated under the same conditions, meet with a number of generations at the end of which the sum total of the modified individuals will surpass that of the unmodified individuals." Giard adds that this law is capable of mathematical demonstration. "Thus the continuity or even the periodicity of action of a primary factor, such, for example, as a variation of the _milieu_, shows us the necessary and sufficient condition under which a variety or species originates without the aid of any secondary factor."