September 17, 1920] 



SCIENCE 



259 



the branching of a tree. It is easy to distin- 

 guish, the twigs and the branches each from 

 each, but where are we to draw the line along 

 each ascending stem? To convey the new 

 conception of change in time we must intro- 

 duce a new set of systematic categories, 

 called grades or series, keeping our old cate- 

 gories of families, orders, and the like for the 

 vertical divisions between the branches. Thus, 

 many crinoids with pinnulate arms arose from 

 others in which the arms were non-pinnulate. 

 We can not place them in an order by them- 

 selves, because the ancestors belonged to two 

 or three orders. We must keep them in the 

 same orders as their respective ancestors, but 

 distinguish a grade Pinnata from a grade Im- 

 pinnata. 



This sounds fairly simple, and for the larger 

 groups so it is. But when we consider the 

 genus, we are raet with the difficulty that many 

 of our existing genera represent grades of 

 structure affecting a number of species, and 

 several of those species can be traced back 

 through previous grades. This has long been 

 recognized, but I take a modern instance from 

 H. F. Osborn's "Equidae":^ 



The line between such species as Miohippiis 

 (MesoMppus) meteulophus and M. irachystylus of 

 the Leptauehenia zone and M. {MesoMppus) inter- 

 medius of the Protoceras zone is purely arbitrary. 

 It is obvious that members of more than one 

 phylum [i. e., lineage] are passing from one genus 

 into the next, and Mesohippits meteulopJius and M. 

 ir achy stylus may with equal consistency be re- 

 ferred to Miohippus. 



The problem is reduced to its simplest ele- 

 ments in the following scheme: 



Our genera are equivalent to the forms of 

 letters: Italics, Eoman, Greek, and so forth. 

 The successive species are the letters them- 

 selves. Are we to make each species a genus? 

 Or would it not be better to confess that here, 

 as in the case of many larger groups, our basis 



2 1918, Mem. Amer. Mus. N. E., N. S., II., 51. 



of classification is wrong? For the paleontol- 

 ogist, at any rate, the lineage a, A, a, a, is the 

 all-important concept. Between these forms 

 he finds every gradation ; but between a and h 

 he perceives no connection. 



In the old classification the vertical divisions 

 either were arbitrary, or were gaps due to ig- 

 norance. We are gradually substituting a 

 classification in which the vertical divisions 

 are based on knowledge, and the horizontal 

 divisions, though in some degree arbitrary, 

 often coincide with relatively sudden or 

 physiologically important changes of form. 



This brings us to the last point of contrast. 

 Our definitions can no longer have the rigid 

 character emphasized by Huxley. They are 

 no longer purely descriptive. When it de- 

 volved on me to draw up a definition of the 

 great Echinoderma, a definition that should 

 include all the fossils, I found that scarcely a 

 character given in the text-books could cer- 

 tainly be predicated of every member of the 

 group. The answer to the question, " What is 

 an Echinoderm?" (and you may substitute 

 Mollusc, or Vertebrate, or what name you 

 please) has to be of this nature: An Echino- 

 derm is an animal descended from an ancestor 

 possessed of such-and-such characters differen- 

 tiating it from other animal forms, and it still 

 ■retains the imprint of that ancestor, though 

 modified and obscured in various ways accord- 

 ing to the class, order, family, and genus to 

 which it belongs. The definitions given by 

 Professor Charles Schuchert in his classifica- 

 tion of the Brachiopoda^ represent an interest- 

 ing attempt to put these principles into prac- 

 tise. The Family Porambonitidae, for in- 

 stance, is thus defined: 



Derived (out of Syntrophiidee) , progressive, 

 semirostrate Pentamerids, with the deltidia and 

 chilidia vanishing more and more in time. Spon- 

 dylia and cruralia present, but the former tends to 

 thicken and unite with the ventral valve. 



■ The old form of diagnosis was per genus et 

 differentiam. The new form is per proavum 

 et modificationem. 



Even the conception of our fundamental 



3 1913, Eastman's "Zittel." 



