High and low Dimoiphism. 323 



a group of this kind, a relation of such a character tliat on the 

 whole the larger species should exhibit more highly differeutiated 

 sexual characters in proportion to their size than the smaller species. 

 In other words in such a gTOup of closely allied species, there 

 should exist, besides high and low individuals within the limits of 

 the species, also high and low species, if the principle of high and 

 low dimorphism has exerted an influence ou the evolution of the 

 group. Now this is exactly what is found in no questionable manner 

 in the groups which I bave examined, in ali of which the pheno- 

 menon of high and low dimorphism within the limits of the species 

 is also apparent. These groups are the Tanaidae, Lamellicornia and 

 Pectinicornia. We shall show that in these groups various series 

 can be traced from species with larger males which exhibit the 

 phenomenon of high and low dimorphism, to smaller and smaller 

 species in which the secondary sexual characters of the males become 

 more and more reduced, so that these smaller species are to the 

 larger what low males are to the high males within the limits of 

 a single species. It is not our purpose to maintain that this factor 

 has been the essential factor in the process of specific dififerentiation, 

 because in this process there are two possibly separate factors, 

 firstly the factor of structural modification and secondly the factor 

 of sexual iucompatibility which prevents fertile intercrossing, but we 

 can maintain with certainty that the principle of high and low 

 dimorphism has played an important part in the structural modi- 

 fication of the group. The argument followed is one of phylogenetic 

 rcconstruction, and this is always hypothetical, but our advantage is 

 this that it is indifferent to us which way the series leads whether 

 from high to low or low to high, and indeed it does not matter 

 whether the actual series we trace is a true one; ali we require is 

 a certain quantitative ratio between the size of the male and- the 

 degree of development of the sexual characters of the various closely 

 related species in a group, and when this is proved our point is 

 proved also. The method employed is necessarily a tedious one, as 

 it must consist in the careful comparison of closely related forms. 



A. Tanaidae. I bave spent some time in working out the 

 Tanaidae of the Bay of Naples and my results are given in detail 

 in the appendix to which the curious reader may be referred. 



Within the family of the Tanaidae certain genera stand apart 

 such as Tanais, Alaofanais, Anarthrura and others: but there is a 

 compact group of very closely allied species which do not exhibit 



