INSTITUTE OF SCIENCE, PHILADELPHIA. 505 



locking giant cilia near their ventral edges. This form offers one step on the 

 way to a gill intermediate between the foliobranchiate and reticulate types, 

 which have hitherto been separated by a considerable gap. A similar step in 

 the filibranchiate direction is offered by the abyssal Callocardia, which most 

 unexpectedly proves to have a gill wholly distinct from its shallow-water rela- 

 tive Isocardia. In Callocardia the greatly-elongated and reflected ribbon-like 

 plates on either side of the stem are fleshy and free, as in Nucula, without any 

 trace of reticulation, but are united by a slender band of connective tissue 

 (possibly carrying a blood vessel, though none could be detected) at their tips. 



If the strap-like plates of Callocardia were narrowed until they became 

 rod-like and the distal fibre omitted, there would be hardly anything except 

 the chitinous skeleton of the filibranch to separate the two types of ctenidia. 

 The gap between the gill of Nucula and that of Pecten or Area, has hitherto 

 been very wide, but the discoveries in regard to Eitciroa and Callocardia (cf. 

 IVoc. U. S. Nat. Mus., vol. xvii, 1895) show that the theory which re- 

 garded them as specializations of a single type was not ill-founded. The dis- 

 tribution of these intermediate or subfoliobranchiate gills is not confined to a 

 single order, but they appear in abyssal members of two different orders ; and 

 their nearest shallow-water allies have reticulate gills. This shows, in the 

 opinion of the writer, that systems based on a single character, whether gills, 

 siphons, muscles or what not, are bound to prove unsatisfactory as our knowl- 

 edge of intermediate types advances ; and that almost any group may have 

 among its members some which retain archaicisms longer than the rest. In 

 such cases the persistency of these characteristics should not oblige us to 

 ignore relationship indicated by other features of the animal. Any permanent 

 classification must necessarily be eclectic, considering all characters and dis- 

 tinguishing sufficiently between genetic and adaptive features. The Nuculoid 

 type of gill has been named Foliobranchiate or Protobranchiate, for the inter- 

 mediate forms above mentioned the designation of Subfoliobranchiate may be 

 used. 



The filamentous gill nominally comprises two series of descending or 

 direct filaments, one on each side of the stem carrying the blood vessel. These 

 may be straight and not reflected, but more usually descend some distance and 

 are then reflected upward externally. The filaments are usually stiffened by 

 minute chitinous rods, but in some abyssal species are soft and perfectly flexi- 

 ble in any direction. To this general type the name of Filibranchiate has been 

 applied. It is confined to the Prionodesmacea. though not all Prionodesmacea 

 are filibranchiate. In the typical ctenidium of this type, exclusive of the stem, 

 the filaments are connected with those adjacent to them only by interlocking 

 cilia disposed in patches or upon the enlarged distal ends of the filaments. 

 Aiwmia, Area and Diinya are examples. 



The next step is the connection of the successive filaments by vascular 

 branches or connective tissue forming a reticulate gill. This, once initiated, 



