THE AMERICAN NATURALIST. [Vol. XXXVII. 



in advance the whole appendage swinging backwards and down- 

 wards. In performing a complete stroke, the apex of the 

 appendage would describe either an ellipse or a fig. oo, while at 

 the beginning and end of each forward and backward stroke the 

 appendage as a whole would rotate on its own long axis, through 

 an angle of something like 45 degrees. These movements would 

 cause the dorsal surface of the basal plate to strike, and rub 

 against the folded margin of the dorsal shield, and would pro- 

 duce just such a series of grooves as are shown so clearly on 

 the plate. These grooves are well worth careful consideration. 



There are three grooves arranged in the arc of a circle just 

 back of the band of coarse pores. On the left side of the plate, 

 Fig. 4, e.g. 1 the groove has a symmetrically rounded floor, is per- 

 haps a fifth of a millimeter deep, and narrows at either end till 

 it disappears. The anterior boundary of the groove is more 

 sharply curved than the posterior. The groove on the opposite 

 side of the plate is less sharply marked, shallower than the first, 

 and its anterior and posterior boundaries are alike. A very nar- 

 row hardly visible groove continues the lines of the first two so 

 as to complete the arc of a circle, but it does not open directly 

 into either of the two marginal grooves. These three grooves 

 are apparently formed by friction against sonic hard object dur- 

 ing the same movement, except that the contact seems to have 

 been more forcible at the beginning and at the close of the move- 

 ment. It will also be observed that these grooves do not lie on 

 the most protruding part of the s id face but in a kind of recess 

 between the most prominent portion and the pitted anterior 

 border, Fig. 6 A. It is clear therefore that the grooves must 

 have been formed by contact with some narrow projecting rim 

 and not by a broad flat surface. This is in harmony with our 

 supposition that the grooves were formed by the rotation of the 

 basal plate of the appendage against either the ventral margin 

 of the cephalic shield or against the adjacent oral plates on the 

 opposite side of the opening to which we have assumed the 

 appendage was attached. 



If one constructs a rough working model of an appendage 

 such as I have described, it will be clear that as the appendage 

 approaches the beginning or end of a stroke, it becomes more 



