384 bulletin: museum of comparative zoology. 



" Thaumantias " of Mayer, and Melicertum, form a natural family, 

 distinguished from all other Leptomedusae by having neither otoeysts 

 nor cordyli, and by the simple radial canals. The name Melicertidae 

 is available for it, though in a narrower sense than Mayer (1910, p. 

 197, "Melicertinae") has used it. 



Should Cyclocanna prove to have otoeysts, it would fall, without 

 difficulty, into the Eucopidae. In any case, it is easily separable from 

 all known Leptomedusae by the peculiar course of the radial canals. 

 The depth and method of its capture, as well as dark pigmentation 

 of its manubrium, suggests that it belongs to the ever growing list 

 of Medusae whose habitat is the intermediate depths, not at the sur- 

 face. 



Cyclocanna welshi, sp. nov. 

 Plate 3, fig. 2-5. 



Grampus Station 10,385, 37° 28' N. Lat., 74° 25' W. Long., August 

 13, 1916, 140-0 meters, 2 specimens, 68 mm. (Type), and 49 mm. in 

 diameter. 



This Medusa is so peculiar in appearance, owing to its canals, and 

 deeply pigmented manubrium, that it can not be confused with any 

 other species. 



In both specimens the bell is of the typical leptomedusan form, 

 saucer shaped, with shallow cavity, the gelatinous substance moder- 

 ately thick near the center, diminishing toward the margin. Both 

 exumbrella and subumbrella surfaces are smooth (Plate 3, fig. 2). 



The course of the radial canals affords the most distincti\'e feature 

 of the genus. These arise from the apex of the manubrium in the 

 usual manner, and are not branched; but instead of running direct 

 to the margin, as is the common rule among Medusae with simple 

 canals, they follow a peculiar S-curve, bending sharply to the left, 

 i. e., anti-clockwise, at about the mid-breadth of the bell, thence run- 

 ning roughly parallel to the margin, almost to the fine of the neighbor- 

 ing perradius, where they once more turn toward the margin (Plate 3, 

 fig. 2). In the type-specimen the outer angle of each canal is close to 

 the inner angle of its next neighbor; but apparently this extreme condi- 

 tion is not attained till maturity, for in the smaller specimen the 

 junction of the radial canals with the circular canal takes place, 

 roughly, in the interradii, though the canals are looped on the same 

 plan as in the type. 



