THE SPONGES. 11 



e.g. pinuli of Il/jaloncma bianchoratum, p. 20. The character of a spicule 



even in minute details may apparently become fixed for the species. Thus 

 in a specimen of H//alonema ovuliferum the rays of the micro-oxyhexacts have 

 the same sudden terminal curving exhibited by the corresponding spicules 

 of Schulze's type specimen, although the two sponges were taken 49' of 



latitude apart, pp. 13, 15. A form of spicule which in some sponges 



varies greatly in size, in other species varies but little. Thus in Gdlius 

 perfomtus the sigmata vary only slightly, p. 129, whereas in Ti/lodesma alba 

 they vary greatly, p. 133. ' 



I do not imdertalve a comparative consideration of the geographical dis- 

 tribution of tlie forms making up tlie collection. Such a consideration 

 would demand a knowledge of the actual systematic value to be attached 

 to many species recorded in the literature of sponges. And such knowledge 

 is not to be had at present. In modern sponge literature, e. g. in the 

 two great monographs of Schulze and SoUas (Schulze, 1887 ; Sollas, 1888), 

 the species conceived are, as it seems to me, what H. M. Bernard contends 

 for in his interesting recent discussions (Proc. Cambridge Phil. Soc. Vol. XI. 

 Pt. IV. ; Verhdlg. V. Intern. Zool.-Congress) of the species-question as 

 affecting the method of recording certain data, viz. homogeneous morpho- 

 logical groups. The sponge species are often very homogeneous, because 

 represented by single specimens. That such groups answer always to 

 natural species, as we understand the word when we speak of the human 

 race, Passer domesticus, Littorina Utorea, or other organisms which we know 

 in great number, is not only open to doubt, but is excessively improbable. 

 It is, I suppose, from this latter point of view (the envisaging clearly 

 the a priori probability that sponges in general exhibit those individual 

 and local differences which all species known intimately exhibit) that 

 0. Schmiilt was led to record in literature the existence of such species as 

 Farrea facunda. Perhaps Farrea facunda is a " natural species," but the 

 data at hand make such a statement only a subjective assumption. Or 

 when the distinguished systematist Topsent expresses the opinion (1902, 

 p. 12) that five species of Poecillasira recorded by Sollas probably represent 

 the variations of two or three species, one is justified in saying "perhaps, 

 but the known specimens differ in certain definite respects." Such sub- 

 jective interpretations of differences perhaps always affect the manner in 

 which we record the occurrence of certain morphological peculiarities in 

 association with geographical and bathymetrical site. But whereas once 



