138 TEST CASES [ch. xii 



agree very well with the theory of natural selection, nor with 

 that of gradual adaptation. There is no doubt that as one proceeds 

 up the scale from variety, the divergence of the characters be- 

 comes greater and greater, and upon these latter theories it is a 

 matter of extraordinary difficulty to explain why the destruction 

 of the intermediate forms should proceed in such a way as to 

 leave groups that present divergences that are more and more 

 marked the higher that one goes in the scale, while at the same 

 time they are quite simple divergences, such as ovary uni- or 

 multi-locular, anther opening by slits or by pores, leaves opposite 

 or alternate, and the rest. Nothing but differentiation can at 

 present explain such phenomena* 



TEST CASE XXIII. DIVERGENCE FROM USUAL 

 FAMILY CHARACTERS 



It is a very noteworthy thing, which the selectionists have found 

 so difficult of explanation, that they have had to fly to their 

 usual refuges, that plants that show great divergences from the 

 characters usual in their families occur, not in the small families 

 (relics or failures) but almost only in the large ("successful") 

 ones. We have given an instance from the Rubiaceae on p. 118 

 and the matter is discussed in detail in (74, p. 621). In the large 

 families one would be inclined to expect constancy, for if it were 

 settled by the early ancestors of Papaveraceae, for example, that 

 a hypogynous flower was the best, why did Eschscholtzia adopt 

 a perigynous one? Solanum, by far the largest genus in its family, 

 opens its anthers by pores, while most of the rest open by slits. 

 Is this the generic adaptation that caused Solanum to become so 

 "successful"? One could go on bringing up cases like these, and 

 there is no escape from the conclusion, so far as our present 

 knowledge goes, that characters of all kinds, however important 

 in classification, may be acquired by single genera at any stage, 

 so that their acquisition is evidently easy, and must almost 

 certainlv be due to direct mutation. Like all the other tests this 

 speaks in favour of differentiation. 



TEST CASE XXIV. PARALLEL VARIATION 



A puzzling case, which the natural selection theory can in no way 

 explain, except by the favourite suggestion of "tendencies", is 

 the parallel variation that so often may be seen. A good instance 

 is afforded by the related families of Eriocaulaceae, Centro- 



