208 THE ORIGIN OF SPECIES [pt. ii 



(2) The permanence with which widely dispersed species re- 

 tain their characters, whether naturally or artificially dispersed. 



(3) With comparatively few exceptions, plants are confined 

 within well-marked limits; sporadics (discontinuously dis- 

 tributed) are rare. If they varied indefinitely, sporadic distribu- 

 tion would be the rule. 



(4) A multitude of allied species of plants grow close together 

 wthout any interchange of specific character. 



(5) The individuals that inhabit the circumference of the area 

 occupied by a species are not found passing into other species, 

 but ceasing abruptly... may meet or overlap similar species. 



[(6) A negative argument in favour of distribution from one 

 centre,] 



(7) The species of the lowest orders (now families) are not 

 only the most Avidely distributed, but their specific characters 

 are not modified by the greatest changes of climate. 



(8) The fact that no plant has been acclimated in England 

 \vithin the experience of man." 



A httle consideration will show that these arguments, with 

 the possible exception of the eighth and last, are as sound to-day 

 as Avhen they were written, and all the work and experience of 

 Jordan (62), Johannsen (61), and the many ecological writers 

 of recent years has but added strength to them. But the stronger 

 they become, the greater is the argument in favour of sudden 

 change by miitation. 



The second and third provisos (about fluctuating variation) 

 above given really go together, for we have no evidence that 

 differentiating variations can appear at all, unless so large and 

 sudden that they are really mutations, not connected with the 

 preceding form by infinitesimal stages. Fluctuating or in- 

 finitesimal variation is simply up and down in the same charac- 

 ters; one never finds a leaf varying by imperceptible stages in 

 the direction of a tendril, or of compoundness, or towards a 

 pitcher. 



A great difficulty for the theory of natural selection, though 

 indeed it is no less for any other theory, is to explain the occur- 

 rence of correlated variations. Why, Avhen a plant produces 

 tendrils, or climbing leaves, should its stem at the same time be 

 weak and flexible? Yet the one would be useless, if not dis- 

 advantageous, without the other. It often happens, in these 

 correlated characters, that while one confers advantage, the 

 other is disadvantageous. It is not altogether wise or reasonable 

 to talk about advantage as having determined progress in nature. 

 To take the single instance of Coleus elongatus (p. 151), its two 



