RELAPSE AND RETROGRESSION. 83 



Ligulates. In the second place, it would be equally- 

 hard to understand why the most primitive 

 Corymbifcrs, such as Eupatoriiiniy should have 

 purple or white florets, while the more developed 

 genera, such as Aster and ChrysantJieniiini, should 

 have most of them yellow. The following hypothetical 

 explanation is suggested as a possible way out of this 



difficulty. Y 



The primitive ancestral composite had reached the 

 stage of blue or purple flowers while it was still at a 

 level of development corresponding to that of the 

 scabious or the Jasione. The universality of such 

 colours among the closely allied Dipsacecu, Valeraincce, 

 Lobeliacece^ and Caiiipamilaccce, adds strength to tliis 

 supposition. The central and most primitive group 

 of composites, the Cynaroids, has kept up the 

 original colouration to the present day ; it includes 

 most of the largest forms, such as the artichoke, and 

 it depends most of any for fertilisation upon the 

 higher insects. Very few of its members have very 

 small florets. All our British species (except the 

 degenerate Carlind) are purple, sometimes reverting 

 to pale pink or white, while Centaurca cyauus, our 

 most advanced representative of the tribe, rises even 

 to bright blue. 



Next to the Cynaroids in order of development 

 come the Corymbifers, some of which have begun to 

 develop outer ligulate rays. ?Iere the least evolved 

 type, Eupatoriuin, with few and relatively large 

 florets, is usually purple or white, never yellow. But 

 as the florets grew smaller, and began to bid for the 

 favour of many miscellaneous small insects, reversion 

 to yellow became general. In a few cases here and 



G 2 



