208 



ing lo the same group servo lo emphasize the distinctness of this larval 

 type. 



This larval "family" appears to be characterized in I lie lirsl stage 

 through the skeleton foiining no basket-structure; the fairly elongate body 

 rod divides at the end into two rather long horizontal branches. In the 

 second stage a posterior transverse rod is formed, from which a pair of 

 short, branched posterolateral rods may issue; the latter may. however, 

 be lacking. There are four vibratile epaulets, but no vibralile lobes. 



The rest of the ('.anuuodonta was arranged by the present author in 

 the families hLchinida>, Toxopneustidtv and Echinometridae, while Clark 

 will not acknowledge the family Toxopneustidae, the forms referred to that 

 family being redistributed in the Echinidse and in a new family Strongylo- 

 centrotidie. The family of the Echinomelrida' is regarded by C.lark as 

 containing only the polyporous. oblong forms, liie oHut, nioi'e primitive 

 forms refiMied to liiis family by the present author being also redistributed 

 by Clark in the Echinidae and Strongylocentrolidae. According lo my 

 view Clark's families Echinida^ and Strongylocentrotidae are quite hetero- 

 geneous, while his family Echinometrida> is quite natural, only, in my 

 opinion, not wide enough. We shall see now what the larva will teach us 

 about this matter. 



It may lirst be stated that, as regards the various genera and the species 

 referred to them, Clark and I in general agree, only with regard to the 

 genus Slrongylocentrotus we disagree in some important points, as is stated 

 below. 



To the family luhinida Clark refers the genera Psammechinus, Lyle- 

 chiniis. Echinus, P<tiechinus, Niidechinus, Evvchinus, Toxopneuslcs, Tri- 

 pneiisles and (iymnecliinus; according lo my view only Psammcchiniis, 

 Echinus and Parcchinus belong lo this family, Evechinus being an Echino- 

 metrid, while the rest of them belong to the Toxopneustida. No larvae 

 belonging lo the gewer a Parechin us (in liie sense of Clark, which I adopt), 

 Xudcchinus or Gijmncchinus are known, while of the other genera we know 

 quite a fair number of larva, so that we can see whether they favour 

 Clark's views or those of the present author. 



The two species known of the genus Psammechinus, miliaris and micro- 

 tuberculaius, have both been studied as regards their development, milidris 

 being very completely known, while there is no description of the fully 

 formed larva of micmlubercuhilus. The larva is characterized by having, 

 in the first stage, long body rods, widening in the end; no basket structure. 

 In the second stage the larva has four epaulets, situated at the base of 

 the four main arms. There is no posterior transverse rod or posterolateral 

 processes. 



