2^60 BULI.ETIN 76, UNITED STATES NATIONAL MUSEUM. 



In 190G (Starfishes of the Hiiwaiian Ishinds) I accepted Verrill's classification, 

 but separated Pfseudarch aster and its alUes as a separate family, hoping to reduce 

 the Goiiiasteridaj to a more homogeneous assemblage. The objection to tliis course 

 is the presence of a pretty complete series of intermediate forms wliich render the 

 adequate definition of faraiUes extremely difficult. 



It would be very convenient to recognize the following families: the Mimas- 

 teridas {Mimaster, Gephyr easier); the Pseudarchasterida; ^nth subfamihes Pseud- 

 archasterinfe (Pseudarchaster, AphrodUaste.r, Paragonaster) and Mediasterinje 

 (Mediaster, Ceramaster, Nereidaster, NympJiaster, Rosaster), and possibly Nectriinse; 

 the GoniasteridaB Math subfamihes Goniasterinae, Chitonasterinte, Leptogonasterinae, 

 and Ilippasterinfe; the Antheneidic {Anihenea, Pseudoreaster). To do this would 

 also necessitate the admission that there are no definite family boundaries. For 

 instance, Pseudarchaster has been found to possess incipient bivalved pedicellarise 

 (Ps. dissonus). Pseudarchaster also has definite superambulacral ossicles and 

 pa.xilhform abactinal plates. Its near relative Paragonaster has (in P. formosus) 

 rudimentary superambulacrals and less elevated tabula to the abactinal plates. 

 The structure of tiie mouth plates of Gephyrcaster alhes that form to Pseudarchaster, 

 while the adambulacral armature is not that of the Pseudarchasterinae, but of the 

 Mimasterinae ; the genus is nearly intermediate. Among Mediaster and its allies, 

 Rosaster (with rudimentary superambulacrals) approaches pretty doselj' the struc- 

 ture of Paragonaster (with the exception of the pedicellariae), wliile Mediaster itself 

 (with rudimentary superambulacrals) is almost inchstinguishable superficially from 

 some forms of Ceramaster. The last group has usuall}^ been called either Pentago- 

 naster or Tosia, and is really intermediate between the Mediasterinae and Goniasterinae. 

 Its species have cither liigh or low abactinal tabula, and no trace of superam- 

 bulacrals. They are closely alhed to perfectly typical GoniasterinaB, such as 

 Tosia (restricted), Plinthaster, and Penfagonaster. The line between a restricted 

 family Goniasteridae and Pseudarchasteridae would be difficult to draw, although 

 the extremes are ver\^ different. In all genera the actinal surface is much more 

 uniform than the abactinal, and the structure of the actinal plates is practically 

 the same. 



The fact that Tlippasteria is usually classed in the Antheneidae (although accord- 

 ing to Viguier, Verrill, and the writer it is closer to typical Goniasteridae) indicates 

 in a measure that no hard and fast line can be drawn between a restricted Gonias- 

 teridiB and Antheneidae. Cladaster Verrill has the abactinal surface exactly hke 

 that of the Goniasterinae, while the actinal is more nearly hke Ilippasteria. Clad- 

 aster, in fact, is intermediate and may be classed in either group. The only feature 

 in which the Antheneida> might be said to differ from the Goniasteridae is m the 

 possession of a thick abactinal skin (wliich is, however, evident^ homologous 

 with the thinner skin of Anthenoides, Leptogonaster, or Stellaster). For the stellato- 

 reticulate structure of the abactinal skeleton is similar to that of Hippasteria (but 

 Cryptopeltaster, classed by Ludwig as Ilippasteria, does not have the secondary 

 abactinal plates in the form of coimecting ossicles, and Cladaster Verrill does not 

 have even secondary abactinal jilates). The pedicellarife are similar in the two 

 groups, but other genera not belonging in either have them {Anthenoides, Gilbert- 



