Occasional Papers of tJie Museum of Zoology 3 



of these authors have united Uranidea with Cottus, as the only 

 character known to distinguish these nominal genera is the 

 number of pelvic rays (i, 4 in Cottus; i, 3 in Uranidea). 

 Snyder^ has called attention to the occasional loss of the fourth 

 soft pelvic ray of Cottus punctulatus. Similarly in Cottus 

 franklinii {q. v.) the fourth ray, although normally lacking, 

 is developed as a rare variation. This lack of constancy in the 

 number of pelvic rays, involving an intraspecific variation 

 which is identical with that used in distinguishing Uranidea 

 from Cottus, indicates that this group is not a natural unit, 

 the fourth soft pelvic ray having probably been independently 

 lost in several cases. Confirming this supposition, the species 

 having but three soft pelvic rays occur in three separated 

 regions : from New England to Labrador and the Great Lakes, 

 in the streams of northwestern United States, and in Japan. 



Key to thk Species of Cottus Inhabiting the 

 Waters of Michigan 



a?' — Preoperciilar spine short and little curved ; head broadly rounded 

 in outline; lateral line normally terminating below base of second 

 dorsal. 



b.^ — Pelvic rays constantly i, 4; palatine teeth normally present. 

 c^ — Distance from anus to tip of snout when measured 

 backward extending to a point nearer base than end of 

 caudal; body rather robust; dark bars definitely de- 

 veloped nieridionalis. 



c.^ — Distance from anus to tip of snout when measured 

 backward extending to a point nearer end than base 

 of caudal; body rather slenderer; dark bars less dis- 

 tinctly developed bairdii. 



b.^ — Pelvic rays usually i, 3; no palatine teeth; distance from 

 anus to tip of snout, when measured backward, extending 

 to a point nearer base than edge of caudal ; dark bars of body 

 not very distinctly developed. 



'Bull. U. S. Bur. Fish., 37, 1907 (1908), p. loi and footnote. 



