A. E. Verrill — Revision Genera and Species of Starfishes. 199 



here that everj^ one of these seven characters fails in certain cases, 

 •and that nearly all of them may occur in each family, so that there 

 is no certain means of deciding in which family certain genera 

 should be placed. Perrier, himself, admits something of the kind, 

 but holds that the preponderance of the characters ought to deter- 

 mine the family in each ease. 



The recent discovery of new genera has so increased the excep- 

 tional cases, by revealing forms that are more or less completely 

 intermediate between the two groups, that it has become difficult to 

 define them in any satisfactory manner. 



The two principal writers who have recently discussed the classi- 

 fication of these starfishes, Sladen and Perrier, have differed con- 

 siderably as to the limits and characters of each group. Thus 

 Sladen included the Odontasteridae (as Gnathaster) and the genera 

 Mimaster and Leptogonaster in the Pentagonasteridse, but Perrier 

 put all these in the Archasteridae. On the other hand, Sladen puts 

 Pseudar chaster and Aphroditaster in the Archasteridte, but Perrier 

 transfers them to the Pentagonasteridae. 



These are well-known genera that have been thoroughly studied 

 by both writers, therefore we must conclude that the two so-called 

 families are not really well defined, natural groups, otherwise such 

 able investigators could hardly disagree to such an extent. 



This question would be of less importance were it not for the fact 

 that in the more general classification of Perrier, these two " fami- 

 lies " belong to two distinct orders. The Archasterida?. he places in 

 the order Paxillosa (op, cit., pp. 28, 29) ; the PentagonasteridDe in 

 the order Valvata. 



The fact that the two so-called families run together, without 

 definite limitations, would necessarily imply that these two " orders " 

 are also badly limited or unnatural groups. 



Almost the only special character by which the two groups can be 

 distinguished, as limited by Perrier, will be the character of the 

 pedicellariae, which, however, are often lacking in both groups. 



But the papilliform pedicellarioe of the Paxillosa, with two to four 

 or more valves, apparently formed from modified spinules or gran- 

 ules, are also found in the Valvata. Sometimes such pedicellaria3 

 are found associated with larger valvular pedicellarire on the same 

 specimen, in the genus Nymphaster and other genera, Avhile well- 

 formed, though small, bivalve pedicellariaj often occur on certain of 

 the antarctic Gnathasterinae, and on other species referred to Paxil- 

 losa. 



