SPONGES 57 



Sollas that the edges of the collars became united by concrescence, giving 

 rise to a continuous membrane, perforated for the passage of the flagella 

 (cf. Fig. 54). Eecent researches have failed to confirm these statements 

 (cf. Vosmaer, Pekelharing, and Bidder), and the appearances seen by Sollas 

 are attributed to defective preservation. The matter cannot yet be con- 

 sidered as settled satisfactorily. 1 



Before leaving the subject of the collar cells, it is necessary to mention 

 the frequently alleged transformation of collar cells and their subsequent 

 immigration into the parenchyma to recruit the ranks of other classes of 

 cells. Bidder (1891) formerly asserted the origin of porocytes in Ascons 

 from modification of collar cells, but this view is now hardly tenable in 

 view of the recent investigations which put the origin of the porocytes 

 from the dermal epithelium beyond a doubt (cf. Minchin [17]). More 

 recently Masterman (1894) has asserted that collar cells when full fed 

 become amoeboid and pass into the parenchyma as trophocytes (see below, 



Flc:. 54. 



Choanocytes with coalesced collars (Sollas's membrane), after Sollas. A, longitudinal 

 section through two flagellated chambers of Anthustra communis, Soil. ; B, diagram of the 

 fenestrated membrane produced by coalescence of the collars, i, prosopyles ; c, aphodi ; e, ex- 

 current canal ; m, Sollas's membrane. 



p. 58), and that further, after having distributed their nutriment to the 

 parenchymal cells, they take up waste products and migrate to the surface 

 of the body, where they act as nephrocytes. It seems more than probable 

 that these statements are founded on mistaken observations. 



(5) The Archaeocytes represent in many ways the most important 

 cell layer of the sponge, but at the same time the one Avhich, up to 

 the present, has been least studied. They are in their nature uu- 

 specialised cells, scarcely modified in structure from the blastomeres 

 of the ovum, and capable of giving rise again, as sexual cells, to the 

 whole organism or, in the gemmules, to any form of tissue (cf. 

 Maas [12]). They stand, therefore, in sharp contrast to the tissue 

 cells, which, having assumed definite morphological characteristics 

 correlated with the performance of particular functions, are only 

 capable of multiplying to form other cells like themselves. The 



1 Numeimis descriptions and figures of collar cells have been published by Lenden- 

 feld at various times, but it is not necessary to refer further to them here. 



