TETRAXONIDA. 45 



evident from the fact that iln- axial i-anal can l>o seen traversing the granular zone. 

 Some <>f the exotyles have one or more swellings in the length of the slmt't . 



The single example was dredged near Winter Quarters, at No. 6 hole, in 130 fins. 



FAMILY HAPLOSCLERIDAE TOPSKNT. 

 SUB-FAMILY GELLIINAE RIDLEY AND DBNDY. 



GELLIUS RODIS. 



(Plate XVII., figs. 1, IA, and Plate XXIV., fig. 1 a-b.) 

 1902. 0W/IM rufu, Topscnt (29. p. 14, PI. I., flg. 9, and PI. III., fig. 4). 



The type of this species, dcscril>ed by Topsent, was n small globular sj>e<-imen 

 27 mm. in diameter. The present collection contains six pieces of this sponge, some of 

 them of large size ; accordingly 1 am enabled to slightly supplement the original 

 description. The largest fragment, 18 x 5 x 2*5 cm. in dimensions, is masnively 

 lobate and closely resembles Petrosia similis var. massa R. and D. (15. plate III., fig. <>).* 

 The other specimens are in the form of thick sub-cylindrical unbranched or branched 

 fragments, the largest (PI. XVII., fig. 1) being 14 cm. in length and 2*5 cm. in 

 diameter. Several of the pieces have oncules about 7 mm. in diameter with circular 

 raised rims. 



The surface, where the dermal membrane is intact, is to the naked eye quite 

 smooth ; though, where the dermis is rubbed off, the surface appears very finely hispid 

 as Topsent describes, and feels rough to the touch. 



A vertical section of a branch in the plane of the axis shows main longitudinal 

 lines of fibres curving outwards from the axis to the surface as they pass upwards, and 

 dividing in a more or less paniculate fashion near the surface ; the secondary fibres, one 

 or two spicules thick in the centre of the specimen, become much thicker a little below 

 the dermal membrane. 



The oxeas (XXIV. la), 403 x 20 /, are sub-tornote. The C-shaped sigmata have 

 a uniform curve almost in one plane : they are 37 M long, 19*4 M broad, and *8 M thick, 



In this variety the skeletal arrangement it very similar to that found in Grlliui rutlii ; the terminal 

 branchlets of the main fibre* are spread out beneath the dermal layer, and anastomose with branchleU from 

 neighbouring main fibres ; this sub-dermal reticuluro, then, is due solely to the spreading out of the periphery of 

 the paniculate ends of the main fibre* (as in O. rtniit), and is not as one might at first suppose a special dermal 

 skeleton ; the central tuft* of the main fibre* proceed vertically up to the surface ; also, not a few scattered oxea* 

 pas* obliquely or vertically upwards from the subdermal network of spicule fibre* ; accordingly the dermal 

 membrane is to a great extent supported on the tips of vertical oxeas. Though Ridley and Dendy clearly saw that 

 the dermal reticulation was only the uppermost layer of the main skeleton (16. p. 11), it was not quite correct to 

 state that the dermal membrane was not supported on tuft* of spicules ; a vertical section shows that it is sup- 

 ported in this manner, viz., by the central tuft* of all the main fibres, as well a* by oxea* from the subdermal 

 reticulnm. The gpicule* of P. timilU var. manu resemble those of O. rudit in shape, but are smaller (vis. 289 x 

 9'76 p). Accordingly P. rimilit var. matta is closely similar to O. rudit in form, in skeletal arrangement, and 

 in spknlation ; in fact it very probably represents an example of a Qrlliiu which ha* lost it* sigmata. I have 

 gone into this matter in some detail because it is always interesting to trace the probable line of descent of 

 Benierine sponge from some form with microsclere*. Further, an instance of this kind furnishes MM support to 

 part of Dendy'* recent scheme of classification, vie., that of making the Gelliinae the starting-point of the 



