THE GERM-PLASM THEORY 



subject in the course of their development to different modifying 

 influences. But this is not the case, a, least not to the extent that 

 this supposition would necessitate. Can it be supposed that, 

 instance, the jumping legs of the water-flea (Gammarus) are a neo 

 sary consequence of the somewhat divergent form of the segments 

 from which they growl A direct proof to the contrary may be found 

 m 'Heterotopia/ for in the place where a posterior limb, modified 

 for holding the eggs, normally occurs in the crab an ordinary walking 

 leg may exceptionally develop (Fig. 90, Bethe), or an appenda 

 resembling an antenna may take the place of an extirpated 

 (Herbst). But if there were really only on.- determinant in the 

 germ-plasm for all the appendages these would of aecessity be 



Fig. 90. The Common Shore-Crab (Carcinus mamas'", seen from below, with 

 the abdomen forced back. In place of the swimmeret, which ought to be borne 

 by the fifth abdominal swimmeret, a walking leg has grown on the lefl 

 and one which properly should belong to the right side o . 1 5, thoracic 

 limbs. 2>s 1-4, swimmerets of the right side. s6, s 7, posterior segments of 

 the abdomen. After Bethe. 



all alike, apart from the larger or smaller differences which might 

 be stamped upon them by growing from segments different in 

 size and in nutrition. Such differences, however, are far Prom being 

 sufficient to explain the great deviations seen among the appendag 

 of most kinds of Crustaceans, and still less to explain their adaptation 

 to quite different functions. 



It need not be imagined that my argument can be controverted 

 by saying that one appendage-determinant in the -vim may split 

 itself in the course of development into a series of different appendag 

 determinants. The question would then arise, How is it able to do sol 

 And the answer can be no other than that the single first determinant 

 had within it several different kinds of elements, which subsequently 



