CAPRELLIDAE OF WESTERN NORTH ATLANTIC 5 



presence or absence of serrations on the grasping margins of the 

 dactyliis and propodus. In Paracaprella both margins are serrate 

 (fig. 41h) whereas in contrast, those of Hemiaegina are completely- 

 smooth (fig. 30b). 



The shape and ornamentation of gnathopod 2 has long been used 

 as a specific character in the Amphipoda, and it is needless to comment 

 on it here except that one must take into consideration the degree of 

 variabiHty this appendage shows at different growth stages. Mayer 

 used the term "poison tooth" to refer to the large tooth on the palm 

 of the propodus. There is e\ddence that glandular material is present 

 near this tooth, and it appears to be venomous in nature (Wetzel, 

 1932, p. 3S7). I have used the term grasping spine when the major 

 "tooth" of the propodus is a spine and have restricted the use of 

 poison tooth to an eminence which is not dehmited at its base or 

 which has previously been designated a poison tooth. Usually grasping 

 spines occur in pairs and when closed the tip of the dactylus fits 

 between them. These spines are found on the gnathopods and the 

 pereopods. 



The niunber of gill pairs was used by Mayer as a generic character. 

 Undoubtedly tliis is an important character but perhaps too much 

 value is placed on it since the gills show various stages of reduction. 

 Some genera uith 3 pairs of gills show a closer relation to genera 

 with 2 pairs than to other genera with 3 pairs, as for example Dodecas 

 and Dodecasella. 



The pereopods on pereonites 3-5 are reduced in many caprelUd 

 genera. Although the number of articles of these rudimentary append- 

 ages is presently important for generic identification, it is often 

 difficult to count the articles, particularly when the terminal article 

 is small or shows some degree of fusion with the penultimate article. 

 Since these appendages are rudimentary and show all degrees of 

 reduction, their value as a generic character is questionable. In 

 Minjerella redunca (p. 75) a female has 2 articles in pereopod 5 instead 

 of the usual 3 and there can be no question that this specimen belongs 

 to tliis species. The use of these pereopods as a generic character 

 seems to mask the phylogenetic relationships of the genera, and it is 

 my belief that the mouthparts provide a better concept of relation- 

 shi[)s. It should be noted that I follow the system of naming the 

 pereopods according to the pereonites upon which they occur; i.e., 

 j)ere<)i)od 3 occurs on pereonite 3. This practice has not been consist- 

 ently followed in the past; various authors (Barnard, Briggs, Guiler, 

 and Huntsman) preferred to number these appendages beginning 

 with pereonite 3. 



The abdomen presents one of the most difficult characters to use 

 for identification. It is extremely small and hard to illustrate accu- 



