308 BULLETIN 82, UNITED STATES NATIONAL MUSEUM. 



Similar cirri, very numerous and very closely crowded, are found on a deep 

 hemispherical or conical centrodorsal in: 



Perometra. Hathrometra. 



Hypalometra. Nanometra. 



ErytJirometra. Eeliometra. 



Trichometra. Solanometra. 



PromacJwcrinus. 



Similar cirri, arranged in ten well separated columns, are found on a conical 

 or columnar centrodorsal which shows extensive radial resorption in: 

 Pterometra. Stenometra. 



Asterometra. Stiremetra. 



Thalassometra (part). Cosmiometra (part). 



Stylometra. Zenometra (part). 



Grotalometra. Balanometra. 



Adelometra. 



Similar cirri, arranged in fifteen columns, which are segregated into radial 

 groups of three columns each, are found in : 



Zenometra (part). 



Similar cirri arranged in fifteen crowded columns on a large thick-discoidal 

 centrodorsal with no radial resorption, are found in: 



Ptilometra. PalseocomateEa. 



The short stout type of cirrus, as seen in Tropiometra, Catoptometra, Eudio- 

 crinus, and in the Charitometridse is found with the same five types of centrodorsal 

 as the long and spinous, though the frequency of the various combinations is dif- 

 ferent, the emphasis being on the first and fifth combinations instead of on the 

 first and second. 



All the other types of cirri occur only on the surface of centrodorsals which 

 range from discoidal to hemispherical or conical, with no differentiation into radial 

 areas, and may be in from one to six or even more rows, alternating, very closely 

 crowded, or with each socket more or less isolated. In general, very slender cirri 

 are numerous and very closely crowded, while stouter cirri are fewer and more 

 scattered ; with slender cirri also the centrodorsal is larger and more hemispherical 

 or conical in shape; but this is due to the fact that slender cirri are only found 

 among the macrophreate forms in which this type of centrodorsal prevails. 



In the smaller groups, such as families or subfamilies, the combination of a 

 certain cirrus type with a particular type of centrodorsal is always of the greatest 

 importance in defining genera, and often also in defining species. 



If we based our deductions upon the study of the comatulids alone, reasoning 

 from the most complex to the most generalized, we should certainly arrive at the 

 conclusion that the cirri of the comatulids were at first five in number, just as we 

 find them to-day five in number in the very young and in the nodals of the penta- 

 crinites, and that each of the five cirri arose beneath the center of the corresponding 

 radial. At the same time we should suppose that the postradial series of ossicles 

 consisted of a linear series, so that the primitive comatulid would be pictured as a 



