REPORT ON THE RADIOLARIA. Ixxix 



(PI. 133, fig. Tl), in whicli the radial spines are absent, not primitively, however, but 

 in consequence of degeneration ; for the twenty cross-shaped perspinal pores, originally 

 due to the twenty radial spines, are still present. In the most nearly allied genera, 

 Porocapsa (PL 133, fig. 7) and Cannocapsa (PI. 133, fig. 8), the proximal part of the 

 twenty radial spines is still present, while their distal portion has degenerated; hence in 

 this case they do not stand in direct communication witli the spherical shell. On the 

 other hand, this primitive connection persists in the genera Astrocapsa (PI. 133, figs. 

 9, 10), and Sphmrocapsa (PL 135, figs. 6-10). The five genera just mentioned form 

 the peculiar family Sphserocapsida (pp. 795-802); the spherical shell is in these cases 

 composed of very numerous small plates disposed like a pavement, each plate or aglet 

 being perforated by a pore canal; in addition to which there are twenty larger (perspinal) 

 pores (or twenty cross-shaped groups each of four aspinal pores) at those important points 

 where primitively the twenty radial spines penetrate the calymma. This peculiar porous 

 "pavement shell" has probably been developed (independently of the twenty radial 

 spines) upon the calymma of the Acanthonida {Acanthoma, p. 749) by the action of 

 the sarcodictyum ; it has, therefore, quite a different morphological significance from the 

 spherical lattice-shell of the Dorataspida, which is composed of tangential apophyses 

 of the twenty Acanthonid spines (pp. 802-847, Pis. 134-138). Each radial spine 

 here forms either two oj)posite or four crossed transverse processes, and since their 

 branches spread over the surface of the spherical calymma and are united suturally at 

 their extremities, the peculiar lattice-sphere of the Dorataspida arises. This extensive 

 family is again divided into two subfamilies: — the Diporaspida (Pis. 137, 138) possess 

 always only two opposite apophyses, and form by the union of their branches two oppo- 

 site primary apertures or aspinal meshes. The Tessaraspida, on the other hand (Pis. 

 135, 138), have always four crossed transverse processes, and form by their union four 

 primary aspinal meshes. From the Dij)oraspida are probably to be derived the Phracto- 

 peltida (p. 847, PL 133, figs. 1-6), the only Acantharia which possess a double 

 lattice-sphere; their double concentric spherical shell may be compared with that of the 

 Dyosphserida. 



119. The Lattice-Spheres of the PhtEodaria. — The lattice-spheres or Sjihgeroid skeletons 

 of the Ph^odaria, which are generally developed quite regularly, though occasionally in 

 a modified form, fall in the order Phseosphseria into two groups of very different 

 structure, each of which includes two families. The first group {Phceosphmia inarticu- 

 toa) contains the families Orosphserida (Pis. 106, 107) and SagosphjBrida (PL 108); 

 the lattice-work of the former consists of irregular polygonal meshes and very coarse, 

 partially hollow trabeculse; in the latter, on the other hand, it consists of triangular 

 meshes and very slender filiform trabeculae; in both families the whole sphaeroid skeleton 

 forms a single unsegmented piece as in most S p h se r o i d e a. In the second group of 



