which debouch into wider, more or less longitudinal canals, in order to terminate in a few 

 main excurrent canals, opening by a simple proct, situated in one of the grooves (PI. II, fig. 5). 

 We are, however, bound to say that we are not quite sure whether the canal represented in 

 fig. 5 is reallv an excurrent one. In the large number of transverse, longitudinal and tangential 

 sections we prepared, we could find but very seldom anything resembling an excurrent 

 apparatus, neither was there seen on the sponge surface any larger aperture, which mighl 

 belong to the excurrent system. We have several times made sections through apertures we 

 observed on the surface of the sponge and which might be excurrent openings. Almost always 

 these turned out to be of quite another nature ; generally they were holes of animals living in 

 the sponge. One of the most probable main excurrent canals opening on the surface is the 

 one figured on PI. II. There was nothing against the supposition. The specimens at our 

 disposal — like those of previous authors — are generally so much contracted that the grooves 

 appear simply as yellowish lines between the red cortical plates. As stated above, the 

 specimen 1033 (PI. I, fig. 5) was not so much contracted; here we could study the incurrent canal 

 system, and, once acquainted with it, we could easily understand the sections made cf contracted 

 specimens. Unfortunately specimen 1033 did not show anything like excurrent apertures. That the 

 system of canals, described above, reallv belongs to the incurrent one, we have a right to 

 suppose because of the abundance of apertures with which it begins. Series of sections prove 

 that the incurrent canals intercommunicate over large regions ; since we found other canals 

 between them not communicating directly with the former ones, we concluded the latter to be 

 excurrent. The narrow ramifications of both are situated near each other, but they do not fuse. 

 Though in such a case we could not actually state it, we must conclude, that the somewhat 

 elongated mastichorions with which prosodi and aphodi are seen in contact, really clo com- 

 municate with the two systems of canals, just mentioned. 



Keller says (1891 p. 325): "Da Rindenporen fehlen, so dienen die schlitzförmigen 

 Oeffnungen der Kanten zur Einfuhr und Ausfuhr des Wassers. Unter jeder Platte befindet sich 

 ein weiter Raum von vierseitig-prismatischer Gestalt, welcher die Halfte des Schwammkörpers 

 beansprucht. Ob er als riesiger Subdermalraum oder als Gastralraum zu deuten ist, muss ich 

 unentschieden lassen, für die erstere Auffassung spricht der Umstand, dass die schlitzförmigen 

 Oeffnungen nicht direkt in denselben einmünden." We cannot agree with these statements. 

 Obviously Keller's specimens were much contracted. What he calls "schlitzförmige Oeffnungen" 

 are nothing but folds of the contracted grooves. Consequently the true incurrent apertures 

 ("Rindenporen") are not absent, but situated at the bottom of the fold. As to the "weiter Raum" 

 we believe this to be a hole, which has been occupied by some commensalistic anima], of 

 which our specimens at least are crowded. We found especially Cirripedia (Balanusr) and 

 Polychaeta. See, e. g., in fig. 2 (PI. II) the large hole on the left hand side. Whether such 

 holes are made by the parasite or not, we are not prepared to say. In many cases Polychaeta 

 certainly occupy canals; this was e. g. the case in the widened portions of the canals marked 

 i. c. in figs. 2 and 6 on PI. II. Why those canals are so much wider, must remain an open 

 question. Is the canal widened by the animal living in it, or could the canal not contract so 

 much as in other places, through the presence of the parasite: 



