890 The American Naturalist. [November, 
natural boundaries. Neglect of these indications has led to 
much of the difference of opinion in the question of geograph- - 
ical distribution, which have been founded principally on the 
conditions presented by the birds and mammalia. 
In this system fragments of existing or old continents, which 
have been subjected to conditions unfavorable to particular 
forms of life otherwise prevalent in them, are, as in the system 
of Sclater, disregarded. Thus, islands generally are not re- 
garded as presenting conditions definitive of divisions of the 
first rank, as was done by Huxley and Gill in the case of New 
Zealand, and Gill and Lydekker in the Polynesian Islands. The 
temperate regions of Africa and South America are certainly 
not separable from the tropical portions, as divisions of primary 
rank, as was done by Allen, who is followed as to South America 
by Gill. With equal propriety western North America might 
be separated from Mississippi and Atlantic North America, on 
account of the great deficiency of its fish fauna. In estimating 
faunistic affinities one has to give similarities over a given 
area more weight than differences, where the differences are 
only due to absence of types. 
Finally, it must be remembered that there are geographic 
points of transition between all the realms. 
I. THE ArcTOGEAN REALM. 
This realm includes three regions, viz. : the Indian, the Hol- 
arctic and the Medicolumbian. I have already defined the 
first two in general terms. The third is the transitional of 
Heilprin, the Sonoran of Merriam and Lydekker, and the 
Neotemperate of Townsend. It embraces what is left of the 
Nearctic of Sclater after the subtraction of the Holarctic. As 
the name Sonoran has been previously given by me to one of 
the districts of this region, I have preferred to use for it the 
mame given by Blanford. 
The faunal characteristics of these regions may be enumer- 
ated af follows: 
Indian Region.—Presence of Holostomatous fishes. Absence 
of Ginglymodous and Halecomorphous and Salmonid fishes. 
Presence of Cæœciliid Batrachia. Absence of Trachystomatous, 
