1/8 THE ARCTOG^EIC REALM. [CHAP. IV. 



Madagascar cannot be excluded. Still, it is quite probable that 

 during the later Tertiary epoch the Ethiopian and Malagasy 

 regions were almost as distinct from the rest of Arctogaea as are 

 Neogaea and Notogaea at the present day, and if such conditions 

 had continued, the former areas would have been entitled to con- 

 stitute a realm by themselves. 



In the sequel we shall discuss the whole number of existing 

 genera of non-volant terrestrial mammals common to the eastern 

 and western halves of the Holarctic region, and likewise such 

 living and extinct types as are respectively peculiar to Eastern and 

 Western Arctogsea. These, taken in conjunction with the fore- 

 going tables, will show that, in spite of the forms common to the 

 two latter areas, there has always been a large number of types 

 restricted to one side or the other of the Atlantic basin. And 

 this leads to the conclusion that, although during a considerable 

 portion, or the whole of the Tertiary period, there was a free land- 

 communication between North America and Eastern Asia by way 

 of Bering Strait, yet that this connecting land must have been 

 comparatively narrow, so that the faunas of the more southern 

 portions of both areas developed to a great extent independently 

 of one another. 



Not the least curious feature in connection with the com- 

 munity of types on the two sides of the Atlantic is the precise 

 parallelism in the development of many of the groups in both 

 areas. In both, for instance, the phyla of the horses and rhino- 

 ceroses were practically similar, although it is thought that the 

 stage occupied in the one area by Hipparion was held in the other 

 by Protohippus. If this particular suggestion should prove to be 

 well founded, it will be self-evident that the true horses have been 

 independently evolved in the two areas ; and it almost seems as 

 if the same had been the case with the rhinoceroses and certain 

 other groups. Had the culminating forms been devolved in only 

 one of the two areas, we should not expect to find the whole of 

 the ancestral links present in both. 



