1908] The Polymorphism of Ariis 57 



At a very early period the ants and social wasps must have 

 made a further advance when the mother insect succeeded in sur- 

 viving till after her progeny had completed their development. 

 This seems to have led naturally to a stage in which the young 

 females remained with their mother and reared their progeny in 

 the parental nest, thus constituting a colony of a number of simi- 

 lar females with a common and indiscriminate interest in the 

 brood. This colony, after growing to a certain size, became un- 

 stable in the same way as any aggregate of like units, and must 

 soon have shown a differentiation of its members into two classes, 

 one of the individuals devoted to reproduction and another class 

 devoted to alimentation and protection. . In this division of labor 

 only the latter class underwent important somatic modification 

 and specialization, while the former retained its primitive and 

 more generalized characters. It is more than probable, as I shall 

 attempt to show in the sequel, that this differentiation was mani- 

 fested in the sphere of instinct long before it assumed a morphologi- 

 cal expression. The social w^asps and the bumble-bees are still 

 in this stage of sociogeny. The ants, however, have specialized 

 and refined on these conditions till they have not only a single 

 marked alimentative and protective caste without wings and lack- 

 ing many other female characters, but in some species two dis- 

 tinct castes with a corresponding further division of labor. In 

 the phylogeny as well as the ontogeny these characters appear as a 

 result of nutricial castration. 



If the foregoing considerations be granted the biogenetic law 

 may be said to hold good in the sociogeny of the ants, for the 

 actual ontogenetic development of their colonies conforms not 

 only to the purely conjectural requirements of phylogeny but 

 also to the stages represented by the various extant groups of 

 social insects. It is clear that we cannot include the honey-bee 

 among these groups, since this insect is demonstrably so aberrant 

 that it is difficult to compare it with the other social insects. 



Comparison of the different genera and sub-families of ants 

 among themselves shows that some of them have retained a very 

 primitive social organization, and with it a relatively incomplete 

 polymorphism, whereas others have a much more highly develop- 

 ed social life and a greater dift'erentiation of the castes. Such a 

 comparison, coupled with a study of the natural relationships of 

 the various genera as displayed in structure, suggests that 

 the advance from generalized to highly specialized societies 



