504 PROCEEDINGS OF THE NATIONAL MUSEUM vol. 83 



include the first caudal segment and count the first compound segment 

 as one, the given count of the trunk segments should be reduced by 

 one to make it comparable with the uniform method adopted herein. 



In occasional specimens the last trunk segment is incomplete, or it 

 would perhaps be more correct to state that it is asymmetrical, since 

 the ventrolateral ridge extends to that segment on one side only, that 

 ridge ending on the preceding segment on the opposite side. Conse- 

 quently, the counts on the opposite sides will differ by one, if the 

 method of countmg outlined here is followed. In such cases the in- 

 complete last segment is uniformly included in the count, and the 

 number of such variant specimens is listed in parentheses in the 

 diagnoses. 



There is one possible important source of error in counting the 

 segments of the trunk as well as those of the tail. The transverse 

 ridge lies along the center of the segment and does not form the 

 boundary between two segments. The natural impulse is to take the 

 space between two transverse ridges to represent a "ring" or segment, 

 but when this method is followed the trunk may erroneously be 

 determined to have one segment less than the actual number. 



How to determine the number of caudal segments. — The first caudal 

 segment differs in shape with the species. In most species it has three 

 spurs or points of intersection of transverse and longitudinal ridges, 

 and thus appears to be hexangular in cross sections, except as an 

 infrequent individual variation, while the following segments are 

 quadrangular. In one species nearly alwaj^s, and in the majority 

 of specimens of another species, the first caudal segment appears to 

 be quadrangular in cross section like the following segments. Bearing 

 this difference in mind, one m.ay determine accurately the number of 

 segments in the anterior part of the tail by counting the spurs on the 

 upper ridge, as in the case of the trunk. However, since the spurs 

 and ridges fade out more or less in the posterior part of the tail, an 

 accurate count of the entire number of tail segments by this method 

 is impossible. The method finally adopted for practical purposes 

 depends on a peculiarity of preserved specunens — a comparatively 

 deep transverse groove usually present on the ventral side, marking 

 the boundary between two segments. The posterior tail segments, 

 therefore, are determined most readily by counting the spaces between 

 the transverse grooves on the lower surface, starting with the last 

 segment and counting forward. The last two segments sometimes 

 have no groove between them, especially in tlie small species, and 

 should be examined with care. If what appears to be the last seg- 

 ment is considerably longer than the preceding one, it most likely 

 consists of two segments; but if subequal to, or shorter than, the pre- 

 ceding, it is most likely a single segment. This may be checked by 



