III. COTTOIDS OF NORTH AMERICA. 27 



into close contact with the prosencephala, sometimes partly overlapped by the latter 

 (in Coitus viscosus, gracilis, and Acanthocottus variabilw) and sometimes entirely ex- 

 posed (in Coitus gobioides, and A. virginianus) . The proportional development of 

 the prosencephalon and mesencephalon varies in both Cottus and Acanthocottus; 

 the former, always larger than the latter, acquires a greater proportion in C. viscosus 

 and gracilis than in C. gobioides, and in A. variabilis than in A. virginiantis. 



Thus, so far, we have no generic difference in the encephala of Cottus and Acan- 

 thocottus. Indeed we have not investigated in that respect the whole range of 

 species, and perhaps characters will hereafter be detected. 



The encephalon of Triglopsis is more characteristic ; its longitudinal axis is pro- 

 portionally greater than in both Cottus and Acanthocottus, a character which we 

 might have anticipated in comparing the structure of the skull and the external 

 appearance of the head. 



The most interesting result which we have derived from the study of the 

 brains of Cotti is the fact that differences are found between the species which we 

 had established, guided chiefly by zoological characters, sometimes most minute. 



If we compare the encephalon of C. viscosus (Fig. 43 45) with that of C. gra- 

 cilis (Fig. 39 41) we find a cerebellum of nearly the same size and form; but the 

 mesencephalic or optic lobes are oblong in the former, and circular in the latter. 

 The prosencephala differ most : irregularly triangular in shape in both species, their 

 surface in C. viscosus, is raised into three nearly equal hillocks, one at each corner. 

 In C. gracilis we perceive only one tubercle-like elevation, situated at the outer 

 and posterior angle, whilst the inner edge of the prosencephalon forms a uniform, 

 longitudinal ridge, the posterior extremity rather projecting beyond and between 

 the mesencephala. 



The rhinencephala in both species are partly covered by the prosencephala, and 

 a little more so in C. viscosus than in C. gracilis. The medial line of separation 

 between these two lobes is rather indistinct on the figure of C. viscosus (Fig. 43). 

 The hypoaria in both species do not differ much, but the hypophysis in C. viscosus 

 is smaller, circular and quite distant from the hypoaria and htematosac, whilst in 

 C. gracilis the hypophysis is oblong, partly incased between the hypoaria and close 

 to the haematosac. 



Behind and between the hypoaria there exists an odd, very small lobe, a little 

 more conspicuous in C. viscosus than in C. gracilis. That lobe does not exist in 

 the two species of Acanthocotti figured on our plate, whilst in Triglopsis it acquires 

 a development greater than the hypoaria themselves. 



In Cottus gobioides (Fig. 42) the prosencephalon is irregularly quadrangular, 

 very uniform above and proportionally smaller than in the two preceding species. 

 The mesencephalon is oblong as in C. viscosus, and the cerebellum circular, more 

 like that of C. gracilis. The rhinencephalic lobes are not completely exposed in 

 advance of the prosencephala. In its general feature it resembles more A. varia- 

 bilis (Fig. 46) than either of the true Cotti. The want of materials, and especially 

 of recent specimens, has prevented our making a more complete study of it as 

 well as of our giving a better illustration. 



