8 Art. 1.— N. Annandale atid T. Kawamura : 



exhalant canals often run horizontally below the dermal membrane 

 for some distance before opening into the osculum, but such 

 channels never form a regular, star-shaped pattern. Spouges of 

 this and the next phases often cover a considerable area on pillars 

 of 1 »ridges and piers and occasionally on the vertical faces of 

 rocks. 



Phase II. (PI. I, fig. 6). This phase, though mainly encrust- 

 ing, is much more massive than Phase I. It often forms relatively 

 thick ramifying horizontal branches. The external parts of the 

 sponge may be tinged with green, but the colour never extends to 

 the interior. The oscula are large, well-defined and round. 



Phase III. (PI. I, fig. 7). The sponges in this phase are among 

 the most remarkable known in the Spongillidae. They form com- 

 pact, ovoid, spherical, irregularly massive or pedunculate masses of 

 a greyish or whitish colour with one or several large oscula opening 

 directly from an exhalant canal of the same diameter. This canal, 

 which forms a more or less cylindrical vertical cavity in that part 

 of the sponge which it drains, is frequently separated into two 

 halves a short distance below the osculum by a thin vertical 

 diaphragm. The external surface is often ornamented with low 

 ridges arranged in diverging groups that interdigitate one with 

 another. The sponge is never more than about 8 cm. in diameter. 

 Sponges of this type are either fixed to hard objects of small size 

 or else lie free on the bottom. 



We will discuss the significance of the three phases in sub- 

 sequent pages. Apart from the external variation, the species 

 may be described as follows: — 



The sponge varies greatly in consistency and may be either 

 soft or hard, in accordance with the closeness or laxity of the 

 skeletal netw T ork. The radiating fibres are always distinct but 

 never massive; they contain very little horny material. The 

 variation in hardness depends largely on the development of the 

 transverse fibres. The oscula are always at least fairly large and 

 the pore-areas distinctly visible with the aid of a hand-lens. The 

 individual dermal pores are relatively large, and the regular dis- 



