56 THE VOYAGE OF H.M.S. CHALLENGER. 



The L i s p li ?e r i d a comprise all those S p h a; r o i d e a iu which the surface of 

 the shell is smooth, without radial spines (Pis. 12, 20). The simplest of these are the 

 Ethmosphserida, with one single lattice-shell, enveloping the spherical central capsule. 

 CenosphcBra, the most simple form of the Ethmosph^rida, may be regarded as the 

 common ancestral form of all S p h ae r o i d e a, in an ontogenetical as well as in a 

 phylogenetical and morphological sense. From this simple lattice sphere all other 

 S p h 33 r o i d e a can be derived either by radial or by tangential growth. If the radial 

 beams, arising from the surface of the simple fenestrated sphere, become connected (at 

 equal distances from the centre) l)y tangential beams, we get the compound shells of 

 the " Liosjjhsei'ida concentrica " (with two, three, four, or more concentric spheres). 

 The radial beams connecting these exhibit in many Liosphaerida the same regular 

 disposition and number as the external radial spines iu the Astrosphgerida. Perhaps 

 these forms in a " natural system " would be better united (e.g., Liosphserida with twelve 

 or twenty internal radial beams, and Astrosphserida with twelve or twenty external radial 

 spines) ; but in many cases (manily for higher numbers) the certain determination of 

 their number and disposition is verj' difficult or quite impossible. 



The Cubosph ffir ida (Pis. 21-25) represent the large and very important 

 family of S p h aj r o i d e a, in which all three dimensive axes are equally distinguished 

 by paii's of spines, corresponding to three axes of a cube or of a regular octahedron, 

 agreeing therefore also with the three axes of the cubic or regular crystalline system. 

 In the majority of the Cubosphserida the six radial spines are accurately op23osite each 

 other in pairs in three axes, perpendicular one to another, and commonly thej" 

 are of equal size and form ; but in some genera the three pairs of spines become 

 differentiated, whilst both spines of each pair remain equal. Either one pair is larger 

 than the two others (which are equal), corresponding to the axes of the cpiadratic crystal- 

 line system ; or all three pairs are difterent (corresponding to the three unequal axes 

 of the rhombic crystalline system) ; the former nearer to the D i s c o i d e a, the latter 

 to the L a r c o i d e a. We may suppose, with some probability, that the Cubosphaerida 

 are for the most part the common ancestral group of those S p h jb r o i d e a, in which a 

 certain number of radial spines or beams is disposed in a regular order ; the Stauro- 

 sphserida may be derived from them by loss of one pair of spines, the Stylosphserida by 

 loss of two pairs of spines, and most Astrosphisrida by multiplying the radial spines, 

 six to fourteen or more secondary sj^ines being added to the six primary " dimensive 

 spines." However, in many Astrosphserida (e.g., in those with eight spines, Centrocuhiis, 

 Octodendron, &c.) the regular geometrical disposition of the radial spines seems to 

 follow another mathematical order, cpiite independent of the Cubosjihferida. 



The Staurosphserida (PL 15) are distinguished bj^ the possession of four radial 

 spines, opposite in paii's in two axes, perpendicular one to another. This rectangular cross 

 determines a certain plane, the " equatorial plane," and this brings the Staurosphserida near 



