14 BULLETIN 82, UNITED STATES NATIONAL MUSEUM. 



specialized and possess division series of 4(3+4) instead of the more primitive 2, 

 character correlations are unstable and uncertain and are liable to sudden and ex- 

 traordinary deviations from the normal, resulting in all sorts of grotesque mixtures, 

 not only within a family or genus but even within a group of specimens of the same 

 species from the same locality. 



Thus, among the highly multibrachiate comasterids individual specimens of a 

 single species may exhibit, more or less well developed, the essential features of 

 entirely different genera from the one to which they really belong. For instance, 

 examples of Capillaster multiradiata are not uncommon with nearly or quite half 

 of their arms of the type characteristic of the species of Comatella, while examples 

 of Comanfhus bennetti are recorded which possess the arm structure of the species 

 of Comanthina and others which possess that of the species of Comantheria; con- 

 versely, specimens of Comanthina schlegelii not infrequently exhibit the arm 

 structure characteristic of Comanihus bennetti. 



This shows the necessity for the utmost caution in determining the genus or 

 species of specimens of the highly multibrachiate forms (especially when some or 

 all of the division series are 4[3 + 4]), and of specimens of 10-armed forms belonging 

 to highly multibrachiate groups. Each individual must be critically examined 

 not only in respect to the essential features of the group as commonly understood, 

 but also in regard to all of the minor features, for it is sometimes found that the 

 character upon which most stress is ordinarily (and properly) laid is in part or 

 even in its entirety replaced by the character normally diagnostic of an entirely 

 different species or even genus. 



The recurrence of nearly or quite identical types of arms, centrodorsals, cirri, 

 pinnules, disks, and other organs in widely different groups raises the question 

 whether such recurrence is really the sporadic reappearance of fixed and definite 

 structural types or whether it may not be merely the result of parallelism. 



Now parallelism is the convergence toward a common type of fundamentally 

 different structures or organs. This convergence progresses far enough to satisfy 

 the requirements of the impelling physical, chemical, mechanical, or economic 

 factors, but no further; hence, though two radically different structures or organs 

 may through parallelism be rendered superficially very similar, the modification 

 is never carried far enough entirely to conceal their ultimate diverse origins. 



In the comatulids identical types of organs and identical structural types, 

 which, as in the case of the method of arm division, are sometimes quite complex, 

 reappear in widely different groups, in each of which they pass through the same 

 developmental history, but in each of which they are associated with other organs 

 and structures of phylogenetically and developmental^ entirely and fundamentally 

 different values which are combined in each case in a radically different way. Such 

 could scarcely be the case were we dealing with structural modifications resulting 

 purely from mechanical, economic, or other exigencies, for we can scarcely imagine 

 parallelisms either to be so erratic in their manifestations and to be in one structure 

 or organ so entirely dissociated from correlated effects upon other structures or 

 organs, or to show, no matter where they appear, the same course of development. 



