154 PROF. ST. GEORGE MIYART ON 



We may now consider those Radiolarian skeletons which were 

 before spoken of as more or less flattened or lens-like capsules, 

 each with a median partition. Such median partition may be 

 taken as representing the radial elements of the skeleton, and the 

 outer shell, of course, as the circumferential portion. 



As before said, the transverse partition extends everywhere to 

 the united margins of the two external halves of each shell ; and 

 thus there comes to be on each side of the median partition a cavity 

 bounded externally by the concave surface of one of the two ex- 

 ternal plates ; these plates are irregularly perforated ; while the 

 internal median partition may consist of a number of concentric 

 rings united together at regular intervals by radiating centrifugal 

 bars of similar size to the concentric rings, as in Trematodiscus. 



These structures may include a spheroidal shell within them, 

 or even three concentric shells of the kind as in Coccodiscus *. 



The shell may be like Trematodiscus , except that it is produced 

 on three sides into three arm-like prolongations, as is the case in 

 Euchitonia ; and these arms may be further united by an exten- 

 sion of similarly formed shell between them, which extension in- 

 creases with age. 



The shell may have four very elongated arms, as in Steplianas- 

 trumf . 



Instead of the median partition being formed of a number of 

 concentric rings, it may be so formed that the parts representing 

 such rings may take the form of a continuous spiral band, start- 

 ing from the outside of a very small centrally placed ring. This 

 is the case in Stylospira. 



Finally, the maximum condition of complexity, with a special 

 arrangement of parts, is attained in Lithelius. To understand 

 this shell, which is spheroidal, or an elliptical spheroid, we must 

 imagine a shell, like that of Stylospira, made so convex on each 

 side that the whole is nearly spherical. Simultaneously with this 

 change the central partition must be widened out on each side 

 so as to fill up the hemispherical vacuity which would other- 

 wise exist on each side of it ; but the continuation of the centri- 

 fugal and circumferential bars must not be imagined to form solid 

 partitions, but to be perforated on each side of each cavity, so as 

 to place all the adjacent chambers in mutual communication. Thus 

 would be produced a set of chambers (all opening into adjacent 



* L. c. pi. xxviii. figs. 11 & 12. 



t See Ehrenberg, Abhand. k. Akad. Berlin, 1875, pi. xxv. fig. 1. 



