GEOGRAPHICAL DISTRIBUTION l8l 



from stream to stream, within their own, countries; and 

 liability to wide dispersal would follow from this capacity 

 as an almost necessary consequence. We can here con- 

 sider only a few cases; of these, some of the most diffi- 

 cult to explain are presented by fish. It was formerly 

 believed that the same fresh-water species never existed 

 on two continents distant from each other. But Dr. 

 Giinther has lately shown that the Galaxias attenuatus 

 inhabits Tasmania, New Zealand, the Falkland Islands, 

 and the mainland of Soutli America. This is a wonder- 

 ful case, and probably indicates dispersal from an Ant- 

 arctic centre during a former warm period. This case, 

 however, is rendered in some degree less surprising by 

 the species of this genus having the power of crossing 

 by some unknown means considerable spaces of open 

 ocean; thus there is one species common to New Zea- 

 land and to the Auckland Islands, though separated by 

 a distance of about 230 miles. On the same continent 

 fresh-water fish often range widely, and as if capri- 

 ciously; for in two adjoining river-systems some of the 

 species may be the same, and some wholly different. 



It is probable that they are occasionally transported 

 by what may be called accidental means. Thus fishes 

 still alive are not very rarely dropped at distant points 

 by whirlwinds; and it is known that the ova retain their 

 vitality for a considerable time after removal from the 

 water. Their dispersal may, however, be mainly attrib- 

 uted to changes in the level of the land within the 

 recent period, causing rivers to flow into each other. 

 Instances, also, could be given of this having occurred 

 during floods, without any change of level. The wide 

 difference of the fish on the opposite sides of most 



