6 R. KIEKPATEICK. 



considerable amount of variation, and will be described separately as A, B, and C. 

 Had tbey not come from the same spot, I would probably have regarded B and C as 

 representing a variety of the typical form. All the specimens possess a well-developed 

 root-tuft, therein differing from those obtained from Kerguelen and other localities in 

 the Southern Indian Ocean, and from near the mouth of the Rio de la Plata in the 

 South Atlantic. Prof. Schulze mentions in the ' Challenger ' Eeport (p. 139) that he 

 had never found the loose root-tuft, which Carter represents in his diagrammatic figure 

 (2, pi. X., fig. 4), and he considers this modification in the original specimen to 

 be conditioned by the looser nature of the substratum. 



Prof. Ijima observes (5, p. 5) that " a case of one and the same Hexactinellid 

 species being firmly fixed when growing on a hard substratum, but producing a root-tuft 

 when living on a soft bottom, has never as yet been shown to exist." The ' Discovery ' 

 and ' Challenger ' specimens of Rossella antarctica seem to me to furnish instances of 

 specimens of the same species being fixed solidly or by a root-tuft in accordance with 

 the nature of the bottom. In spite of the considerable range of variation in bodily 

 form, mode of fixation and spiculation, the Antarctic and the more northerly forms 

 appear to me to come within the limits of one and the same species, but I regard the 

 latter as belonging to a variety — var. solida. The microdiscohexasters, too, are 

 considerably larger in the typical Antarctic specimens than in the northern variety. 

 If the anchor spicule figured by Schulze (6, pi. Iv., fig. 12) really belongs to the 

 specimen from which it was obtained, its presence suggests that the sponge, now solidly 

 fixed without a root-tuft, was derived from a form with such an appendage. 



Specimen A. — (Plate I., fig. l, and Plate IV., fig. 2 a-g.) 



This, which is the smallest of the three specimens, closely resembles the original 

 one obtained by Sir James Ross and diagrammatically figured by Carter (2, pi. x., 

 fig. 4). The total length is 7 cm., the body being 4 cm., the root-tuft 2 cm., and 

 the oral fringe nearly 1 cm. in length. The greatest breadth is 3 cm. The velum 

 is about 7 mm. from the surface of the body. Numerous oval or pyriform buds, each 

 about 2 mm. in length, are present amidst the dense tangle of the velum. The orifice 

 is narrow, oval, about 1 cm. in length, and acute-angled at each end. . The diactin 

 marginalia surrounding it form a sloping palisade which meets in the middle line 

 above, the inclination increasing from without inwards. The root-tuft, which encloses 

 debris of worm-tubes, fragments of Polyzoa, etc., is formed of pentactins in the shape 

 of four-pronged anchors with thick curved prongs. The surface of the body is level, 

 though here and there small, barely perceptible conuli occur. The skeleton is mainly 

 formed of long wavy bundles of slender diactins. 



Spicules.— The principalia are long slender diactins, often with roughened, 

 rounded or bulbous ends. There are also parenchymal regular hcxactins (IV., 2a) 

 scattered about. The marginalia are sharp-pointed oxydiactins 1 • 5 cm. long and 

 160 ju, thick. The basalia are long, slender, four-pronged anchors 4 cm. in length, with 



