j6 SMITHSONIAN INSTITUTION 



morphological changes I have mentioned; but whichever is selected for this 

 purpose, we find our arrangement more or less disturbed by some of the others. 

 For example, a fairly satisfactory arrangement could be based upon the modi- 

 fication of the primary brachials, which would correlate quite well with other 

 characters, if it were not for the fact that this would throw Sagenocrinus and 

 Forbesiocrinus into different families; whereas the connection between these 

 genera is so evident, and' the line of descent so probable, that any scheme which 

 compels their separation is unacceptable. We cannot, of course, represent lines 

 of descent in space of two dimensions, so that anything in the way of a 

 diagram or table would be imperfect, even if we knew all the facts. Still less 

 is it practicable when many of the relationships rest wholly upon conjecture. 



In some groups of the crinoids family divisions are most sharply marked. 

 No one need ever be in doubt, from inspection of the calyx alone, whether a 

 Camerate crinoid belongs to the Rhodocrinidae, Melocrinidae, Actinocrinidae, 

 Batocrinidae, Platycrinidae, or Hexacrinidae. This cannot be said of the 

 Flexibilia. By reason of the fundamental difference in construction of the 

 two groups, there is not in the latter that sharp demarcation between calyx and 

 arms which is so characteristic of most of the Camerata. Here, on the con- 

 trary, in by far the greater number of the genera, the calyx passes into the 

 arms by imperceptible gradation, so that in the fossil state, being usually un- 

 able to see any part of the tegmen, we cannot tell with certainty where the 

 calyx ends and the free arms begin. The different modifications of this struc- 

 ture also shade into one another by various transitions, so that groups of 

 family rank may be formed, as above stated, which differ somewhat according 

 to the character which is taken as the basis of division. 



Nevertheless, it seems possible to form a reasonable opinion as to the 

 relative importance of the characters as the basis for large divisions : 



i. The differentiation of the anal area, being completely developed in the 

 earliest Silurian, and continuing almost to the end of its history, may be taken 

 as marking the most primitive division of the group. It evidently dominated 

 the lines of descent throughout, and should therefore be accorded first impor- 

 tance in the definition of families, all others being subordinate modifications 

 affecting one or the other of these lines, but probably not interrupting them. 



2. The presence or absence of interbrachials affords a useful basis for 

 subordinate divisions. 



3. The differentiation of the brachial system in the number of primary 

 brachials, although evidently affording characters of much importance, is one 

 which has impressed itself with varying force upon the two primitive lines, 

 and not in parallel progression. It may be assumed to be a subordinate modi- 

 fication, marking the limitations of genera, and perhaps of sub-family divisions. 



