CILETOPODA 299 



unknown whether each of the three mature forms is distinct 

 and hereditary, or whether the progeny of one pair of 

 parents may take one of the three different courses according 

 to the conditions of life. In the former case the three forms 

 would be separate varieties of one species if not three 

 distinct species. If the second alternative is correct it is 

 very difficult to suggest any explanation without more 

 complete knowledge of the influence of various conditions 

 of life on the sexual maturity of the worms. It seems 

 probable that the problem in this special case is the same 

 as the general problem why certain species are pelagic and 

 differentiated in the mature condition, while others reproduce 

 without any change of habits or structure. We can scarcely 

 avoid the assumption that the latter is the more primitive 

 condition, and thus we are led to suppose that something in 

 the conditions of life in certain species caused them to 

 produce eggs with less yolk, and to swim about while 

 shedding these eggs. According to my views, this change 

 produced directly the structural metamorphosis in the 

 parents. In Nereis dumerilii such conditions of life, what- 

 ever they are, may be supposed to occur occasionally, but 

 not uniformly, acting on some individuals, but not on all. 

 In that case the peculiarities in the ova and in the structure 

 of the parents may have become hereditary, but only in 

 association with the special conditions of life, just as in the 

 case of the Axolotl. Every individual may be held to have 

 inherited the tendency to metamorphosis in its own structure 

 associated with the production of pelagic ova, but actually 

 to exhibit these phenomena only when a particular stimu- 

 lation calls them forth. In a precisely similar way every 

 sterile female in polymorphic ants, bees, or termites inherits 

 the power of developing into a perfect fertile female, and 

 we know that this is so from the fact that under the proper 

 conditions the development always actually takes place. 



