ASTEROIDEA OF NORTH PACIFIC AND ADJACENT WATERS FISHER. 19 



papulae are used, although in the transitional or intermediate family Linckiidae 

 these characters are variable, which is tantamount to sajnng that the order is not 

 sharplj' defined. This is indeed the case. 



A study of the majority of genera in the Phanerozonia has led me to rearrange 

 somewhat Verrill's revision of Sladen's classification, especially as regards the 

 Astropectinidse. The characters used in determining the families are emphasized 

 in the key. These relate to the water vascular system, to the modifications of 

 the abactinal skeleton, to the structure of the alimentary canal, to the modifica- 

 tions of the marginal plates, and structure of actinal intermediate plates. The 

 pedicellariae are not used, except in connection with several primary characters. 

 They are suggestive for determining the relationships of species and genera, but 

 are unreliable for any higher groups in this order. The superambulacral plates 

 are important when taken in connection with other characters; their absence is 

 often an important character in determining kinship. The gonads are chieflj' of 

 use in determining relationship of species. Their arrangement is apparently 

 constant within a genus, but can not be used for families. 



Not all the primary characters are of use for every family, and such characters 

 as the form of the tube feet and abactinal plates are common to several families. 

 They might better be called superfamily or subordinal features. Thus the order 

 may be divided into two groups, those with pointed tube feet (no definite sucking 

 disk being present) and those in which the tube feet have a definite sucking disk. 

 The first group nearly coinci<!es with Perrier's Paxillosa, and includes the families 

 Porcellanasteridse, GoniopectinidaB, Astropectinidae, Luidiida\ The first family 

 is sharply marked off by the possession of only a single ampulla to each tube foot; 

 the other three have two. The first family also has a peculiar modification of the 

 marginal plates, in the form of cribriform organs," and differs from all the others 

 except the Luidiidse in lacking an intestine and intestinal coecum. The Luidiidae 

 are separated from the Goniopectinidise and Astropectinidse by the structure of 

 the alimentary canal just referred to, and by the abortion of the superomarginal 

 plates. The Goniopectinidse constitute an aberrant type near the Astropectinidse, 

 from which they differ in having simple marginal cribriform organs as well as 

 webbed fascioles between the actinal intermediate plates. The family is not 

 large, and holds an intermediate position between the Porcellanasterida? and Astro- 

 pectinidae. But the segregation of the group from the AstropectinitUr is desirable, 

 as it leaves the latter much more homogeneous. The Astropectinidac include all 

 genera with pointed tube feet, double ampulla^, intestine, superambulacral plates, 

 and true paxillie (or the readily derivetl parapaxillffi). The presence or absence 

 of an anus does not figure, as this character is worthless, for it varies within a 

 genus. Thus the family includes Plutonaster and its allies, classified by Siaden, 

 Perrier, Ludwig, Koehler, Gregory, MacBride, and their followers in the Arcliasterida\ 

 and by Verrill in a distinct family (which can not be tliagnosed), the Plutona.steritUv. 



Before considering tlie second group, that characterized by having tube feet 

 with suckers, a word concerning the pointed tube feet is necessary. Several 

 authors have stated that the p()inte<l tube feet are tlirectly related to the habit 

 of living on a muddy bottom, and have in away discredit^'d the character for classi- 



o Present also in the Goniopectinidse, sdthough always of a simple type. 



