2 54 The Geographical Distribution of Fishes 



different than they could be if we imagine an ancient shore 

 connection. If New Zealand and Patagonia were once united 

 other genera than Galaxias would be left to show it. 



3. VCe know nothing of the power of Galaxias to survive 

 submergence in salt water, if carried in a marine current. As 

 already noticed, I found young and old in abundance of the 

 commonest of Japanese fresh-water fishes in the open sea, at 

 a distance from any river. Thus far, this species, the hakone * 

 dace, has not been recorded outside of Japan, but it might well 

 be swept to Korea or China. Two fresh-water fishes of Japanese 

 origin now inhabit the island of Tsushima in the Straits of 

 Korea. 



4. The fresh-water fishes of Polynesia show a remarkably 

 wide distribution and are doubtless carried alive in currents. 

 One river-goby t ranges from Tahiti to the Riu Kiu Islands. 

 Another species, J originally perhaps from Brazil through Mexico, 

 shows an equally broad distribution. 



5. We know that Galaxias with its relatives must have been 

 derived from a marine type. It has no affinity with any of the 

 fresh-water families of either continent, unless it be with the 

 Salmonidffi. The original type of this group was marine, and 

 most of the larger species still live in the sea, ascending streams 

 only to spawn. 



When the investigations of geologists show reason for 

 believing in radical changes in the forms of continents, we 

 may accept their conclusions. That geological evidence exists 

 which seems to favor the existence of a former continent, Ant- 

 arctica, is claimed on high authority. If this becomes well 

 estabhshed we may well explain the distribution of Galaxias 

 with reference to it. But we cannot, on the other hand, regard 

 the anomalous distribution of Galaxias alone constituting proof 

 of shore connection. There can be no doubt that almost every 

 case of anomalies in the distribution of fishes admits of a possi- 

 ble explanation through " the slow action of existing causes." 



Real causes are always simple when they are once known. 

 All anomalies in distribution cease to be such when the facts 

 necessary to understand them are at our disposal. 



* Lcuciscus hakucnsis. f Elcolris fusca. J Awaous gciitvittatus. 



