12 WTM.IS : RVmRNCK ACAtXSr THK ORI(!(N (»K 



U. macropoda, both having red pendulous caq>els, but the 

 former liaving them on stalks I inch long, the latter on stalks 

 6-7 inches long. These two species are otherwise most closely 

 allied, and both grow together. Now, can we suppose that 

 the long stalks of the carpels of U. macropoda are so great 

 an advantage that they should have been developed without 

 suppressing at the same time U. Narum? Or, if preferred, 

 the argument may be put the other way, and it may be asked 

 why U. macropoda was not suppressed. This will get over 

 the difficulty of origin from a common ancestor, the length 

 of the stalks in wliicli may have been intermediate. 



If, instead of taking the endemics at random, we pick those 

 that occupy only particular hill tops, we obtain as striking 

 proofs of our general thesis as that given by Coleus elongatus, 

 but the point seems to be sufficiently made out. 



If the characters that distinguish the endemic species of 

 Ceylon, which number about 800, were characters really 

 valuable to them, we should expect in general that they would 

 be common within the island. On the other hand, they are 

 usually remarkable for their rarity and for being confined to 

 one locality in the forest, or to one mountain top or group of 

 high mountains. Thus, to take at random the first 55 met 

 with in Trimen's " Flora," 37 are very rare, and only 18 at all 

 common ; while of the non-endemic species in the same genera 

 only 8 are rare, while 19 are common. 



Or, again, take the endemic genera. Here, at any rate, the 

 evolution having gone far enough to give rise to genera, one 

 would expect to find commonness the rule, but in actual fact 

 the Ceylon endemic genera are usually rare. 



The Ceylon endemics, then, are in general a group of some 

 800 species, of which about two-thirds or more are rare, 

 confined to one or a very few localities. In particular they 

 affect mountain tops and places in the south-western forests. 

 Every isolated mountain top has some, e.g., Ritigala, the 

 Pedurutalagala group of mountains, Adam's Peak, Hiniduma- 

 kanda, and, as re(!ent investigations made by myself would 



