Classification of the Asconidse. 353 



and scarcely or not at all wider than the surrounding tubes ; 

 but in any ease it has distinctly the appearance of a mere 

 exhalant opening for the system of tubes of which it is the 

 central point. The Ascons of this group correspond very 

 nearly to Hackel's families Tarromida and Nardopsida. 



In the second group, of which hotryoides is the type species, 

 the osculum is large in comparison with the network of tubes, 

 which it completely dominates. The tubes, often very 

 minute, as in hotryoides itself, form a basal or subbasal net- 

 work, from which arise the erect oscula, the latter having 

 much more the appearance of individuals than of simple 

 cloacae. The oscula grow in height and send out radial 

 diverticula, sometimes over their whole surface, which con- 

 tinue to grow and become ramified, forming branches which 

 either may themselves become new oscular tubes, giving rise 

 to an arborescent colony, or may anastomose to form a net- 

 work of tubes, from which new oscula in their turn arise. 

 Thus the adult colony consists usually of two parts — a fine 

 network of tubes, from which arise at intervals the erect 

 oscula, often of great size. This mode of growth is liable to 

 considerable variation, from the grape-like cluster of oscula 

 presented by hotryoides on the one hand to the arborescent 

 '■'■ pinus^'' '^ form of complicata on the other; but the adult 

 colonies of this group can never be confused with the reticu- 

 late masses formed by the coriacea type. 



These differences of form and growtli, so hard to describe 

 accurately but so easy to recognize, are shown to be of generic 

 value by the fact that they occur correlated with a great 

 number of other characters — structural, histological, and 

 embryological — of which I can only mention here the most 

 salient. 



In the first or coriacea group the principal skeletal spicules 

 are equiangular triradiates, some or even the majority of 

 which may become quadriradiate by the addition of a fourth 

 or gastral ray, but without the three basal rays (or, as we 

 may conveniently term them, the triradiate systenis'\) losing 



* Hackel's figure of " Ascandra pimts " gives quite a false notion of 

 tills form, for it does not really stand erect and foursquare to all the winds 

 that blow, as his figure would lead one to believe, but it is a creeping 

 form, attached at numerous points to the algae &c. among which it 

 grows. The oscular tubes only become erect to any considerable degree 

 when it grows in dense clumps, such as are not uncommon in deep water 

 otf the Mewstoue at Plymouth. 



t I shall use the phrase " triradiate systems " in this paper to denote 

 the triradiate spicules or the basal rays of the quadriradiates. While in 

 the quadriradiates the basal rays are formed exactly in the same manner 

 as the triradiates, the fourth or gastral ray has a distinct origin and 

 appears later. 



