January 10, 1900.] 



SCIENCE. 



^y> 



aeeording to the knowledge of their time 

 the groups seemed pretty sharply separable, 

 yet that does not mean that we should take 

 those merging genera as a model for our 

 own work. It is, for instance, a very dif- 

 ferent matter at this time to pronounce the 

 great and long united genus Astragalus a 

 complex of distinct genera, for with our 

 present knowledge of the gradual transi- 

 tions between these alleged segregates of 

 Astragalus, their treatment as separate 

 genera involves a sort of suppression of 

 facts. 



It is for this reason that writers who 

 divide large genera on the ground that the 

 component parts are just as good genera as 

 many which are traditionally maintained 

 are doomed to the disappointment of seeing 

 their propositions neglected. We may 

 readily pardon the older writer who in his 

 limited Imowledge of species made some 

 incorrect inferences as to generic lines, who 

 pointed out the probable channels about 

 his archipelago on the basis of a conscien- 

 tious but necessarily restricted exploration, 

 made, let us say, with a still ineffective 

 sounding apparatus. It may even, for 

 purposes of geographic orientation, be 

 worth while to let his hypothetical channels 

 remain on our charts, their shallowness and 

 danger being duly indicated. It is quite 

 another matter when we are directed by 

 modern writers to record such channels on 

 the chart at places where we know that 

 they do not in reality exist. But it may be 

 asked why, if it is sometimes desirable to 

 leave the record of an old error, may not 

 the very similar new proposals be equally 

 useful for purposes of geographic orien- 

 tation. If it is worth while to let old and 

 now merging genera stand, why not make 

 new ones of the same kind? But this end 

 can be accomplished equally well by the 

 use of subgenera or sections, and that too 

 without our seeming to indicate that differ- 

 ences are more constant than they really 



are, without our being obliged to record a 

 navigable channel where none exists. 



The division of large genera into several 

 smaller ones is commonly accompanied by 

 a certain loss about which little has been 

 said, the obscuring, namely, of the exist- 

 ence of the larger affinity which had been 

 indicated by the old and more comprehen- 

 sive genus. However diverse the elements 

 of CEnothera may be, it can not be doubted 

 that they form a recognizable whole, of 

 which the constituent species are more 

 nearly related to one another than they are 

 to Epilobiicms, Gauras or to species of 

 other onagracious genera. In other words, 

 CEnothera in its comprehensive sense is a 

 ijatural, although perhaps rather loose 

 group. If we show the narrower affinities, 

 by the use of subgenera, we still have the 

 word CEnothera left to cover the larger 

 relationship. This is a matter of impor- 

 tance, for it gives the student at once the. 

 information that the affinities of the parts 

 of the genus are closer than those of the. 

 genus to other genera. If we give up the 

 old genus CEnothera and substitute a group 

 of small component genera, we lose sight 

 of the larger affinity and our classification 

 is accordingly less clear and rich in its 

 statement. It has suffered a distinct loss 

 in the abandonment of the genus CEnothera 

 in the larger sense. An added disadvan- 

 tage is that we are making the constituent 

 groups, although they are of a nearer affin-. 

 ity and clearly belong to a subordinate rank 

 in classification, appear as if they were co- 

 ordinate with other genera which are still 

 treated in the larger way. It is clear, 

 therefore, that the division of a loose but 

 more or less natural genus is attended by 

 some 'disadvantages quite unconnected with 

 any considerations of sentiment or tempo- 

 rary inconvenience. 



We have seen that the difficulties of. 

 classifying plants in a really natural and 

 logical way are somewhat increased by the 



