GENUS SEBASTODES. 577 



one of the parietals reaches the middle line and the other 

 does not; in other cases the posterior part of one parietal 

 and the anterior part of the other reach the middle line, 

 and yet a wide strip of supraoccipital separates the two 

 bones throughout their length. All of these conditions 

 are evident in the accompanying figures. There is no 

 more reason why that condition of the parietals in which 

 they barely meet should be chosen as the line of separa- 

 tion between two groups of species than that any other 

 degree of approximation or overlapping should be chosen. 

 The character is unfitted a priori to serve as a primary 

 character. The kind of difficulties into which its adop- 

 tion leads is illustrated, among other instances, b}^ the 

 fact that S . elongaUis and S . levis are placed in the group 

 with separated parietals, although in some individuals the 

 parietals plainly meet. 



Not only is the condition of the parietals, by the nature 

 of the character, unsuited for the purpose which it is 

 made to serve, but it is not correlated with a single other 

 important cranial character. After it is adopted as the 

 primary character it does not serve in the slightest degree 

 as a key to the rest of the structure. The degree of de- 

 velopment of the cranial spines and ridges, the condition 

 of the interorbital space, the curvature of the base of the 

 skull, the condition of the ventral process of the basi- 

 sphenoid and the direction of the mesethmoid processes 

 are all closely correlated with each other and all lead to 

 the same arrangement of species. The condition of the 

 parietals not being correlated with a single other charac- 

 ter, its use as a primary character is bound to rupture all 

 the correlations that do exist; and that is what it does. 

 To select a single illustration from among a host of them, 

 the genus Sebastomus, as made up by Eigenmann and 

 Beeson, includes species from all parts of the group: 



