310 BULLETIN 100, UNITED STATES NATIONAL MUSEUM. 



appear to be so interconnected as to form a single, continuous sys- 

 tem, connecting at intervals with the endochonal canals. 



In some species there are deviations from the type, in the matter 

 of the ectochrotal canal system, which tend toward a secondary 

 acquisition of larger efferent canals and simple oscula. In G. megas- 

 trella (see Topsent, 19115) for instance, there is a cloaca into 

 which open numerous efferent canals, varying in diameter, each 

 roofed over by a fenestrated membrane. Some of these canals in the 

 neighborhood of the cloacal wall fuse, the common space (larger 

 efferent canal) so formed being roofed over by a large fenestrated 

 area of the cloacal lining. It is not recorded whether excurrent tub- 

 ular endochonal canals are present in this species. The sterrastral 

 layer, Topsent finds, is practically absent from the cloacal wall. A 

 second step in the same direction, complementary of the above, is 

 made by G. perarmata in which Dendy (1905) finds that the cribri- 

 form membrane lining the cloaca includes here and there larger 

 openings " where the pores have apparently become confluent and 

 thus formed small vents. " 



The species such as Geodia barretti Bowerbank (Sollas, 1888, p. 

 250), with uniporal excurrent chones, are referable to Sidonops 

 (Lendenfeld, 1903, p. 101). Some later species with simple sphinc- 

 trate oscula, G. micropunctata Row (1911, p. 296) for instance, are 

 likewise referable to Sidonops. In the case of a number of older spe- 

 cies assigned to Geodia in Sollas' monograph the records leave it in 

 doubt whether excurrent chones exist. The oscula in such forms are 

 described as simple or the implication is that they are simple, and it 

 is quite possible that the efferent canals are of the ordinary type com- 

 mon in tetraxonida. Several of these species have been referred tt 

 Sidonops by Lendenfeld (1903), and, unless a special genus is made 

 for them, this seems to me the best procedure. 



Dendy has recently discussed the limits of Geodia. He has doubts 

 (1916, p. 254) as to the value of the character and arrangement of 

 the incurrent and excurrent orifices for the discrimination of Geodid 

 genera, a set of points on which Sollas and later writers have laid 

 stress. He therefore uses Geodia in a wider sense than is customary, 

 his definition of the genus, relating only to the skeleton, covering 

 Geodia (plus Cydonium) sens, str., I sops Sollas, Sidonops Sollas, 

 and Caminella Lendenfeld. 



The common practice is, it seems to me, still the better one. Prob- 

 ably the genera intergrade, but the extreme types are well marked, 

 and many sponge genera intergrade with respect to any one of the 

 several points that go to make up our conception of a genus. As to 

 the biological value of such differences in the canal system, who 

 knows? Perhaps they are quite as hereditary as many skeletal 



