GEODIA AMPHISTRONGYLA. I77 



ing the cortex radially and broken off at the surface; are probably the proximal 

 parts of rhabdomes of these anatriaenes and mesoclades, which composed the 

 spicule-fur in the Uving sponge. 



The microscleres are oxyasters, oxysphaerasters, small strongylosphaer- 

 asters, and sterrasters. The oxyasters are confined to the choanosome and are 

 not numerous; the oxysphaerasters occur chiefly in the walls of the cortical 

 canals and in the roofs of the subcortical cavities. The small strongylosphaer- 

 asters form a dense layer on the outer surface. The sterrasters occupy the 

 sterraster-armour layer of the cortex in dense masses and are scattered also in 

 large numbers throughout the whole of the choanosome (Plate 20, figs. 33, 39). 

 The choanosomal sterrasters are mostly young forms. 



The large amphistrongyles (Plate 20, figs. 1-3, 17-19) are slightly or con- 

 siderably (Plate 20, figs. 2, 3) curved, 0.5-2.3 mm. long, and, in the middle, 

 18-32 jtt thick. Their thickness is by no means in proportion to their length, 

 the shortest amphistrongyles being quite as thick as or thicker than the longest. 

 The long amphistrongyles (Plate 20, figs. 1-3) usually taper from the centre 

 towards both the rounded ends, the thickness of the latter being 40-75 % of the 

 thickness of the former. This attenuation may be equal at both ends — then 

 amphistrongyles appear as isoactines; or it may be unequal — then the amphi- 

 strongyles appear as anisoactines. The shorter the spicules are, the less marked 

 is the attenuation towards their ends. The spicule represented on Plate 20, fig. 19, 

 which is 1.05 mm. long, is at the ends 17 pL and 20 /x, only in the middle 22 n 

 thick. Still shorter amphistrongyles are, if isoactine, regularly cylindrical or 

 slightly thickened at the ends, or, if anisoactine, club shaped, like the spicule 

 represented on Plate 20, fig. 18, in which the thickest part is situated at one 

 end. Slender amphistrongyles with knob-shaped thickenings (Plate 20, fig. 17), 

 which may be young forms, have been observed occasionally. It is certain 

 that the short, stout amphistrongyles do not develop into the long, slender ones. 

 That the former are cylindrical and often more or less thickened at the ends, 

 while the latter are never thicker and usually more slender and attenuated 

 towards the ends, leads me to suppose that the silicoblasts, which build up the 

 ends of these spicules, wander, while they perform their allotted task, either 

 slowly or rapidly in a distal direction. In the first case short, thick, and more 

 cylindrical, in the second case long and relatively slender amphistrongyles, 

 attenuated towards the ends, are produced. 



The rare large styles (Plate 20, figs. 20, 21) have similar dimensions as the 

 amphistrongyles above described and may be considered as derivates of aniso- 



