ASTROSCLERA W'lLLEYANA, THE TYPE OF A NEW FAMILY OF SPONGES. 465 



which it agrees with that of sponge spicules, and differs from that of Alcyonarian 

 spicules'. 



The scattered linear spicules (spic.) represented in Fig. 8 occur both in the soft 

 parts and embedded in the skeleton. I believe them to be of foreign origin. 



Origin of the skeletal elements. The gelatinous layer which invests the upper surface 

 is crowded with young growing skeletal elements, the small ones free and spherical, the 

 larger packed together like hailstones, and assuming the polyhedral form (Figs. 5, 7, 

 8, and 11). 



The spherules take their origin in single cells of the jelly, near the upper surfiice 

 (Fig. 5, a — d). In the early stages of growth the nucleated granular cell body is seen 

 as a thin investing layer surrounding the spherule, which is from the first composed 

 of radiately arranged crystalline fibres. The smallest spherules that I have recognised 

 are Ifi in diameter, and I am inclined to think, from this fact and from the appearance 

 of the organic basis described above, that they are of this size when they are first 

 formed. As the spherule increases in size it takes up its position as an element of 

 the fixed skeleton, and in the course of their growth the angular spaces between 

 adjacent skeletal elements are completely filled in, to the exclusion of the soft parts. 

 The elements thus lose their spherical shape and become polyhedral. The external 

 surface of a spherule, in contact with the layer of soft tissue, is often beset with 

 radiating points, and resembles a portion of a spheraster of a siliceous sponge (Fig. o, 

 /' and (j). I find no trace of a skeletogenous layer of cells covering the fully formed 

 skeleton ; it appears probable that the cells in which the spherules take their origin, 

 remain in relation with them till their growth is finished, but I have only seen them 

 in the early stages. 



The arrangement of the skeletal canals. Figures B and C are views of an approxi- 

 mately median section through one of the Lifu specimens. The canal system consists of 

 intimately anastomosing channels, the greater number of which have a direction 

 upwards and slightly outwards, opening on the ujiper surface and connected with one 

 another by abundant transverse and obliquely running communications. 



In Figure A 5 a section of the interior of the Funafuti specimen is seen. Near 

 the centre of the cut surface a group of large canals, having a diameter of from 60 

 to 80/x., is cut across, whose arrangement corresponds with that of the large pores seen 

 at the surface of this specimen. The canals run close to one another and their general 

 direction is upwards and outwards, diverging from the axis of the fungiform ' original 

 growth.' In addition to those forming the central groups there are others running 

 approximately parallel with them in planes which are set radially to the axis of the 

 group. Moreover, the section lays open at least one canal which is approximately 

 transverse to this axis, and whose course thus corresponds with those radially directed 

 grooves, the initial stages of canals, seen in the Lifu specimen represented in Fig. 1, a. 

 In the rest of the cut surface smaller canals are exposed. 



It appears probable that the large canals carry the ultimate trunks of the efferent 

 system of the sponge. I have not succeeded in recognising systems of large canals in 



' For this fact, as well as for his assisitance in staining some sections for me with this reagent, I am 

 indebted to Mr G. C. Bourne, Fellow of New College, Oxford. 



