105 



'J'llH AXAriXK (ii:.\rS XVKOCA A.\I> IIS XI'.AUI'.SP ai.i.iios. 



IlAKKV (". OliKUllOl.SKK. 



'IMi.' follnwiiii,' i.ii.LCcs |.n-s('iit lilt' results of ;iii invest i^iitioii. lM■^MIn many 

 ye.ii-s a.i,'(). iiild the iiciierlc i-elat ioiisliips of the iliicUs e(iiiiiii(.iily iiicliMled 

 ill till' .iiciiera Aiimra Fleiniim iMarila aiiet. reel and Fiili'nd'i Stei>lieii.s. 

 Those species are : 



Xifrncd f I rill (I ( I.iiinaons) 



Xjirocii (iiiicriciUKi (Kyloii) 



Mjro'd nilisincrid (Wilson) 



Xi/roni hiicri ( Katlde.l 



Xijrocd inintcd (Giildenstiidt) 



Syrocd iiuiolnta Salvador! 



Ni/roc(i (iH-straHs Eytoii 



Nyroca hnnoiea Eytoii 



Nyi-oc(i' niitioiii (Sclater and Salvin) 



Nyroca cri/throplithaliiKi (Wied) 



Fuliffula marihi (Linnaeus) 



Fuligula affiiiix Eyton 



Ftilif/tila fiiligiilfi (Linnaeus) 



Fulifjula novaescclandiae (Gmelin) 



Fuliguld coUarin (Donovan). 



Of the fifteen species thus eoinpi-ised. we ha\c exaniiiied all hut t wo - 

 Xijroca innotdta and Nyrocd iidtioiii. Study of the structural characti-rs of 

 these birds has resulted in the discovery of some hitherto unnoticed char- 

 acters, which indicate that to include all these species in a single genus 

 \v(Uild be manifestly improper, and that in view of important differences, 

 some additional genera need recognition. The characters liere used as 

 generic-ally diagnostic are constant and trenchant so far as it has been 

 possible for us to verify them, and a number of other characters we have 

 omitted because they fail to come up to this standard. It is becoming fairly 

 well understood that many differences that are apparently good generic 

 characters fail when several individuals of a s])ecies. or when species of 

 other supposed generic gron])s are critically examined: and while a mere 

 ahnormality in a single individual can not he lield as invalidating a gen- 

 ei-ic character, it is ((uite evident that a character which is not i>raclically 

 constant in all individuals of a sjiecies can scarcely he nsahle in the diag- 

 nosis of a genus. It is evidently thus unsafe to base generic characters on 

 examinati(m of a single six'cimen of a species, hut sullicient examples should 

 he exandned to eliminate the factor of individual variation. As in many 

 ducks, so in the liirds at present under consideration, the form, shape, and 

 l>roportions of the bill are of prime importance as tlie indicators of generic 

 relationships. Of other characters made use of in the present conni'ction. 

 tho.se of the relative proportions of the wing and of iinier toe with claw, 

 compared with the exposed culmen. are apparently of most importance. 

 While coloration as a i»rimary geneiic cii;irac(ei- is of little or no value 

 among these ducks, it is of interest to note that the generic classification 



