cviii THE VOYAGE OF H.M.S. CHALLENaER. 



means excluded. For even in the skeletonless primitive genus of all the Spumellaria, 

 Actissa (as well as in the social Collozoum), there are found, in addition to the usual 

 spherical types, other species (or subgenera, p. 12) whose central capsule is not spherical 

 but a modification of the sphere ; in Actiprunum ellipsoidal ; in Actidiscus lenticular ; 

 in Actilarcus lentelliptical ; if such modified forms of Actissa were to develop their 

 lattice-shells independently, then their form would correspond to that of the central 

 capsule ; and such simple ellipsoidal, discoidal, and lentelliptical lattice-shells might 

 have been the primitive forms of the Prunoidea, Discoid ea and Larcoidea. 



164. Genealogical Tree of the Sphceroidea. — CenosphcBra, the simplest form of the 

 spherical lattice-shell, may be unhesitatingly regarded as the common stem-form of all the 

 Sphseroidea (pp. 50-284, Pis. 5-30). CenosphcBra (p. 61, PI. 12) arose directly 

 from Actissa simply by the silicification of the spherical exoplasmatic network of the 

 sarcodict}T.im around the central capsule, on the surface of the concentric calymma. 

 From this simple siliceous extracapsular lattice-sphere all other forms of Sphseroidea 

 have arisen, in the main by the manifold combination of two simple processes, first by 

 the formation of radial spines on the surface of the lattice-sphere, and second, the addition 

 of concentric spherical lattice-shells. Both processes may be utilised as the foundation 

 for a systematic treatment of the Sphgeroidea (compare pp. 52-58). 



If in the Sphseroidea the characteristic number and disposition of the radial spines be 

 regarded as the most important heritable peculiarity of the different families, then we have the 

 following natural arrangement : — (1) Liospha^rida, without radial spines ; (2) Cubosphferida, with 

 six radial spines (opposite in pairs in three axes perpendicular to each other); (3) Staurosphserida, 

 with four radial spines (in two a:ses crossed at right angles) ; (4) Stylosphserida, with two opposite 

 radial spines (in the vertical main axis); and (5) Astrosphterida, with numerous regularly or 

 irregularly distributed radial spines (eight to twenty or more). If, on the contrary, more stress 

 be laid upon the number of the concentric lattice-shells, then we have the following artificial 

 grouping : — (1) Monosphserida, with one simple lattice-sphere : (2) Dyosphaerida, with two concentric 

 lattice-spheres ; (3) Triosphrerida, with three ; (4) Tetrasphserida, with four ; (5) Polysphi^rida, with 

 numerous (five to twenty or more) concentric lattice-shells ; (6) Spongospha?rida, with a spongy 

 spherical shell. In general the former arrangement appears more natural than the latter, since 

 the number of primary radial spines, which grow out from the primary lattice-sphere, determines 

 their ground-form from the outset, whatever may be the number of secondarily added shells. 

 Strictly speaking, according to the view adopted, those Liosphterida which have several shells, 

 on the outer surface of which there are no radial spines, ought to be classified according to the 

 number and arrangement of their internal radial connecting beams and cUstributed among the 

 other families. The practical application of this correct principle meets, however, with great 

 difficulties. Also in many cases the phylogenetic relations of the different Sphteroidea are 

 more complicated than would appear from both these classificatory principles. In general their 

 phylogeny will quite correspond with their ontogeny, since from the innermost first formed 



