[NTRODUCTK >N. 



th, ipitous cliffs on the other side of the backbone of the 



ret home of the very rare and beautiful .4 chatinclla 



mind was constantly seeking an answer as to why many of the 



I iwaiian genera of snails should have an area of distri- 



ition not more than a mile or two in length, while, in the case of some 



rial niollusks in other parts of the world, the district 



thousand miles or more in length. Again, of Hawaiian 



^, why should those living continuously in the trees, without 



ling to the ground even for breeding, occupy on the average 



much smaller than those occupied by species living contin- 



isly on the ground? 



The mystery was only intensified when I observed a certain corre- 

 lation between the form of the island on which the species had origi- 

 nate 1 and the method of grouping and distribution of the species and 

 varieties. In the first place, the forest species on one island are never 

 mpletely intergraded with those on another island. Again, the 

 inter^rading of nearly allied species on one island usually relates to 

 cies found in contiguous valleys; while the most divergent forms 

 are found in the districts that are most widely separated. On West 

 ai, which is a single conical mountain, deeply furrowed with val- 

 leys and gorges radiating from one center, we find each group of 

 species King in a circle around the mountain, each species occupying 

 it> own district, though intcrgrading with those of adjoining districts, 

 and no one of the species strongly divergent from any of the others 

 of the same group a distribution that seemed symmetrical and 

 impressed me as strikinglv similar to the distribution of groups of 

 birds and mammals around the North Pole; for example, the distribu- 

 tion of species of bears throughout the Northern Hemisphere. But, 

 in strong contrast with this, is the distribution of species of snails on 

 the island of Oahu. Here the forest region, in which the snails are 

 mostly found, is not spread in a circular form over a group of radiat- 

 ing valleys, but lies in a strip about 35 miles in length and from 2 to 

 6 miles in width, upon a mountain range; and the forms of one 

 closely related group are distributed in two parallel series of species 

 on opposite sides of the ridge, the most- divergent forms being those 



* In Plate II, figs. 11-25, are given 1.5 species of Achatinella, distributed in the 

 ■ if live valleys, and, therefore, limited to an area less than 5 miles in 

 length and not more than 2 miles in width. Of these 15 species there are nearly 

 bundred easily distinguished varieties. Moreover, several other much rarer 

 ■i Achatinella found on the vegetation of the same district have been de- 

 scribed as species; and of 6 other genera of the family Achatinellidae 

 there are, within the same limits, 17 or 18 strongly marked species. 



