442 ANALOGICAL RESEMBLANCES, 



that tlie teeth in either case have been adapted for tearing 

 flesh, through the natural selection of successive variations; 

 but if this be admitted in the one case, it is unintelligible 

 to me that it should be denied in the other. I am glad to 

 find that so high an authority as Professor Flower has come 

 to this same conclusion. 



The extraordinary cases given in a former chapter, of 

 widely different fishes possessing electric organs — of widely 

 different insects possessing luminous organs — and of 

 orchids and asclepiads having pollen-masses with viscid 

 discs, come under this same head of analogical resem- 

 blances. But these cases are so wonderful that they were 

 introduced as difficulties or objections to our theory. In 

 all such cases some fundamental difference in the growth 

 or development of the parts, and generally in their matured 

 structure, can be detected. The end gained is the same, 

 but the means, though appearing superficially to be the 

 same, are essentially different. The principle formerly 

 alluded to under the term of ajialogical variation has prob- 

 ably in these cases often come into play; that is, the mem- 

 bers of the same class, although only distantly allied, have 

 inherited so much in common in their constitution, that 

 they are apt to vary under similar exciting causes in a 

 similar manner; and this would obviously aid in the ac- 

 quirement through natural selection of parts or organs, 

 strikingly like each other, independently of their direct 

 inheritance from a common progenitor. 



As species belonging to distinct classes have often been 

 adapted by successive slight modifications to live under 

 nearly similar circumstances — to inhabit, for instance, the 

 three elements of land, air and water — we can perhaps 

 understand how it is that a numerical parallelism has 

 S{ metimes been observed between the subgroups of distinct 

 classes. A naturalist, struck with a parallelism of this 

 nature, by arbitrarily raising or sinking the value of the 

 groups in several classes (and all our experience shows that 

 their valuation is as 3^et arbitrary), could easily extend the 

 parallelism over a wide range; and thus the septenary, 

 quinary, quaternary and ternary classifications have prob- 

 ably arisen. 



There is another and curious class of cases in which 

 close external resem.blance does not depend on adaptation 



