305 



where only a portion of the tarsus was removed, in several of 

 which there appeared to be no defect in the imago. These 

 exceptions are, no doubt, correlated with the determination 

 there seems to be that, whatever happens, the claws and some 

 of the last tarsal joint shall be reproduced, and claws are con- 

 sequently found on most rudimentary limbs that are merely 

 nodules. 



Though the strong tendency of the claws to be reproduced 

 probably modifies the result, there seems reason to believe that 

 the regeneration is more perfect in the case of the part injured 

 than in more distal ones. An injury to the femur, for example, 

 results in the femur being more completely reproduced than 

 the tibia and tarsus, the latter being much less complete than 

 when the tarsus alone had been removed at the same larval 

 stage. 



Another conclusion is that when there is amputation by a 

 clean incision, regeneration takes a simple and straightforward 

 course, but where crushing takes place, and possibly therefore 

 division of the group of embryonal cells that provides for 

 regeneration, there may take place various supplementary 

 portions, branches, and duplication of limbs. All my experiments 

 of this character were made in late (usually last) instars, and 

 consequently the supplementary parts only developed in the 

 crude way that equally occurred, in such instars, in cases of 

 amputation, in the de\'elopment of the limb itself. 



A comparison of the results in Agrotis -prónuba and Li par is 

 dispar shows that the rapidity of regeneration is ver}' much 

 greater in some species than others. 



39 



