ON MERLIA NOEMANI. 671 



A second kind of microsclere, viz. a very minute^ slender, 

 simple sigraa is found in fair abundance in the choanosome, 

 especially in the immediate neighbourhood of the flagellated 

 chambers. At one time I thought these spicules were the 

 broken curved euds of rhaphides, but latterly I have seen the 

 little spicules in their scleroblasts. 



Rhaphides and trichodi-agmata constitute a third and 

 fourth kind of microsclere. 



The Spicules. — The slender tylostyles (PI. 35, fig. 1), 

 which are commonly curved at the distal end, though some- 

 times nearly straight, ai-e about 140 fx long, 1'8 /i thick, and 

 with oval heads 5 by 2'2 fx in length and breadth. The 

 rhaphides (PI. 35, fig. 2a) are about 80 ju long. They are found 

 separately or mixed in with tylostyles. Trichodragmata (PI. 

 35, fig. 2) occur, but are rather rare. In one specimen there 

 are toxa-like spicula with a central kink or bend (PI. 35, fig. 

 13), but this is an exceptional feature. The clavidiscs (PI. 

 35, figs. 3-9) are about 45^ ^'^^g; 30 fx broad, and with the 

 rim, which is bevelled inwards to a thin edge, 3 n broad. A 

 key-hole shaped sinus or notch is present on the inner margin 

 at each end of the long axis. The axial canal is in the centre 

 of the thickness of the outer edge of the rim. Numerous 

 variations and sports occur, which are interesting because 

 they show the mode of origin of these spicules, viz. from 

 deeply curved rods which have bent round till the ends met 

 and joined. Sometimes the ends cross or do not meet at all, 

 or a transverse bar may cross from side to side (PI. 35, figs. 

 7, 8). Fig. 14 shows clavidiscs with a disc-like plate in place 

 of the key-hole sinus. Fig. 7 shows a sigma-shaped spicule 

 which is probably merely a deviation from the ring shape. 

 Again, the key-hole sinuses may be absent from one or both 

 ends. Lastly, the clavidisc 7nay sometimes be in the form, 

 not of a ring, but of a solid disc (not figured). 



I had formerly supposed (6) that the clavidiscs were related 

 to chelate spicules of Desmacidonid^, but I now consider 

 their aflanities to be Avith the diancistra of Hamacantha 

 (see Section 6 on the affinities of Merlia). These spicules 



