58 Journal of the Mitchell Society [December 



skeleton is on the axinellid plan, including an axial dense aggregation 

 of spiciilo-fibres from which radial fibres pass outward. The radial 

 fibres may be comparatively stout, sometimes (expanded and forming 

 lamellae, or may he reduced to one or a few spicules, or may even dis- 

 appear as structures distinct from the dermal brushes in which they 

 terminate. Connectives between the radial fibres generally absent, 

 occurring only in a few species. The dermal brushes characteristically 

 include one or a few long megascleres surrounded by numerous shorter 

 ones, but the entire brush may be represented by one or a bunch of 

 long megascleres, whereas in Axinella the radial fibres terminate "in 

 tufts of diverging spicules slightly smaller than the bulk of the spi- 

 cules" (Vosmaer). Skeletal spicules smooth and characteristically 

 monactinal, but diactinal forms (to be looked on as variants of the 

 type) are frequently intermingled, and in rare cases most of the 

 spicules are diactinal. Acanthostyles characteristically occur, echi- 

 nating the skeletal fibres and scattered, but in some species they are 

 greatly reduced in number, in others absent (subgenus Syringella). 

 Microscleres generally absent but may occur as bundles of rhaphides. 



In considering a generic concept such as that just formulated one 

 must remember that classification is no mere mechanical assorting, a 

 putting of like with like. That much of it is easy and delightful, but 

 the fact that the features used in the classification of a group are or 

 may be independent variables, that is, may occur together or some may 

 be absent or greatly modified, makes classification more dependent on 

 argument than is generally recognized. Often, to be sure, no argu- 

 ment is conclusive, in the end there being several options about equally 

 good. In such a case some consensus of opinion as to the best, most 

 useful, practice is desirable. All this applies to the delimitation of such 

 genera as Raspailia. 



One other consideration must be kept in mind in considering such 

 generic concepts, and that is that variation may have proceeded in 

 such a way as to make it quite impossible, in the present state of our 

 knowledge of diagnostic points, to recognize the relationship of a given 

 species to the genus. If, for instance, the acanthostyles are lost, a 

 typical habitus and dermal skeleton would still show the relationship, 

 or even the peculiar dermal skeleton might be accepted, in the absence 

 of acanthostyles and typical habitus, as a safe index. But the dermal 

 bunches, as we have seen, may also vary from the type, and if habitus, 

 peculiar dermal skeleton, and acanthostyles all go, nothing remains to 

 show the relationship. I take it that such eventualities may have 



