74 THE COLOURS OF FLOWERS. 



less due to reversion, and the blue to progressive 

 modification, not yet fully selected by insects ; so 

 that in all probability it is now actually in course of 

 acquiring a new colour. Much the same thing hap- 

 pens with the common pimpernel (Anagallis arvensis], 

 Its ancestral form is probably the woodland loose- 

 strife (Lysimachia nemorum), for though the capsule 

 of the pimpernel now opens transversely, it still retains 

 the five dark lines which mark the primitive dehis- 

 cence ; and in other respects it most closely resembles 

 Lysimachia^ which is a bright yellow. But pimpernel 

 itself is usually orange-red, while a blue variety is 

 frequent on the continent, and sometimes appears in 

 England as well. Every botanist can add half a 

 dozen equally good instances from his own memory. 



Highly variegated flowers show the greatest ten- 

 dency of any to such occasional reversions, or, as it 

 is usually put, are extremely variable. The pansy 

 ( Viola tricolor) is an excellent example. The snap- 

 dragons, orchids, and irises are also cases in point. 



Indeed, the extent to which flowers are modifiable 

 in the hands of gardeners largely depends upon the 

 amount of modification which they have already 

 undergone in the natural condition. Very highly 

 developed plants have on the one hand acquired a 

 great constitutional plasticity of nature, and on the 

 other hand have a large number of previous stages to 

 fall back upon. Hence gardeners can do almost any- 

 thing they like with Dahlias, Cinerarias, Asters, 

 Chrysanthemums, and other advanced corymbiferous 

 composites; with Calceolarias, Antirrhinums, and 

 other Scrophularinece; with pansies, pea- flowers, heaths, 

 and lilies ; with exotic Gloxinias, Bignonias, Tecomas, 



