GEOGRAPHICAL DISTRIBUTION OF PLANTS 345 



we see it, and which other characters are indifferent in the 

 actual struggle for life. 



In other words, the principle of adaptation, as one of the 

 main parts of the theory of evolution, should be separated 

 from the study of the geographical distribution. This 

 latter science itself should be divided into two parts, one 

 of which would be concerned with the delimitation of the 

 regions inhabited by organisms of various degrees of affinity, 

 while the second would have to explain the directly observed 

 facts of local occurrence and actual migration. The first 

 of these two parts is a comparative science and is directly 

 related to the theory of the common origin of living beings. 

 The second must become an experimental inquiry into the 

 relationship between the qualities of the plants and those 

 of the environment, which it may prefer or endure. All 

 speculations upon the relationship of organisms to special 

 features of this environment which attempt to explain 

 larger groups of -characters on the assumption of some 

 adaptation, are, to my mind, as yet merely poetical descrip- 

 tions of the way in which we should like to understand 

 and admire nature, but not facts capable of direct proof. 



Desert plants afford an instance which may give a 

 clear appreciation of the two contrasting methods of explain- 

 ing the nature of plants. According to the current view 

 they are most astonishingly specialized and adapted for 

 large regions where it is impossible for other plants to 

 thrive. Although belonging to numerous natural families, 

 and therefore showing hardly any genetic affinity among 

 themselves, they enjoy a group of common features w r hich 

 show the closest imaginable relation to their arid environ- 

 ment. Three main types of desert plants may be distinguished. 

 The most common is composed of the low shrubs with green 

 stems and twigs, a loose mode of branching, and small 

 coriaceous leaves or, in some instances, with no foliage at 



