44 AMERICAN HYDROIDS. 



DI/IKIIIK mi should be dismembered, part of the species going to Sei-tiiliii>!. and part to 

 T/iniiii'iii, as above defined. 



For convenience in referring specimens of American Sertularidse to their proper genera, 

 according to the plan adopted in this work, the following key is presented, with the confession 

 that, like all such keys, it is purely artificial and does not indicate the interrelationships of 

 these genera: 



KEY Til TUP: liENKHA OF AMERICAN' SEKTl'LAKID.K. 



Hydrothec-ie nil on one side of the branches, their distal ends being turned alternately to the right and 

 Mt - - Hydrallmania. 



Hydrothec;e arranged on all sides of branches. 



( Iperculum of one adcauline flap, no anastomosis of branches ^I'lui/iim/isix. 



( )])ercnliiin of several flaps, branches freely anastomosing Dirti/n,-/ii</iinn. 



Ilydrotheeie arranged in pairs, each pair being revolved so as to be at right angles to the pair immedi- 

 ately above and below (Slniinilln'i-n. )' 



Hydrothecie always in two longitudinal rows. 



Hydrothccje in groups of pairs, the uppermost being decidedly smaller than the lower 1'iixi/t/n-ir. 



Operculum adcauline, and of one flap. 



Hydrothecal aperture large, body not flask-shaped lli/ilinxin. 



Hydrothecal aperture small, body flask-shaped ...[liii'linnrin. 



Operculum abcauline, of one Hap, or of two or more flaps. 



Operculum of three or four pieces.- Hydrothepe strictly alternate, margin usually toot bed Si r/ /.</. I In. 



Operculum of one (abcauline) flap, or of two, rarely three, 3 flaps. 



IIydrothec;c strictly opposite, each internode normally bearing a single pair Sfftnlnrin. 



Hydrothetve siibopposite to alternate, each internode normally bearing more than two Tliiiim-in. 



Operculum wanting, margin round. 



Branches normally arising from the lumen of a hydrotheca ( Tin I'lii-lniHum. ) 



Gonangia normally arising from the lumen of a hydrotheca Synthedum. 



In concluding this general discussion of the genera of the Sertularidse the author wishes to 

 explain his position in view of some of the more important objections which he apprehends will 

 be urged against the- classification here adopted. 



firxf. It will be said, and truthfully, that the system is based on a heterogeneous .set of 

 characters, and that different characters are used in defining certain genera from those used in 

 defining others. Of course, the ideal system would be to find one or two characters that would 

 suffice. As a matter of fact, no writer could use more care and ability to effect this end than has 

 Lcvinsen, as we have seen; and his work has been chiefly valuable in demonstrating the impossi- 

 bility of such a method, at least so far as the Sertularidge are concerned. There is also a distinct 

 danger in confining diagnostic features within too constricted limits, and this is that it is almost 

 certain to result in a most unnatural assemblage of species into genera which do violence to actual 

 affinities, as was done, for instance, when Ilijili'tiViminin was regarded as a plumularian on account 

 of using the one character of unilateral arrangement of hydrothecse. 



With increase in experience the naturalist sees more and more clearly that all characters 

 must receive due consideration, and that the fewer the characters used the less satisfactory will 

 be his groups, if he desires them to indicate real affinities. 



Again, it has come to tie recognized among systematists that a given character will often be 

 of the greatest value in one section of a family, or other group, while the same character is 

 comparatively worthless in another section of the same family or group. Thus the opereuluni is 

 almost an ideal character to use in separating out the genus J)!J>}I<IK!I(, as used by Levinson, but 

 fails in TJnintria and A'< rhilin-'m. 



Second. A still more serious objection will doubtless be raised by those who will discover 

 iiitergrading forms between the genera that are here recognized. 



'The brackets indicate that the genus is not found in American waters. 



2 Very exceptionally the liydrotliecie in this genus are without teeth, and the operciilum is stretched like a 

 drumhead acmss the very wide, round aperture. 



"When three teeth are present, as in some of the llfxninxri/jiliiix group of Hi'rtnlnriit, the upper one is much 

 smaller than the others, and the Operculum is very delicate, while the three-toothed forms of ,S'i iinlitn:l/n have equal 

 and equidistant or conspicuous teeth, and the operculum is evident. 



