SCIENCE. 



[N. S. Vor,. XXIII. No. 577- 



guishing traits are constantly associated. 

 Wlien limits are properly drawn it is cer- 

 tainly true that a very large number of 

 such ideal genera exist. Unfortunately for 

 the peace of mind of the systematist, how- 

 ever, there are considerable series of species 

 in certain families, which quite defy classi- 

 fication into genera of this sort. They are 

 groups in which we are forced into accept- 

 ing a far less satisfactory type of generic 

 division, and in some cases it is necessary 

 to make the most of a single character. Of 

 this fact, the genera Arenaria and Stellaria 

 furnish an excellent example. So far as I 

 am aware there is no technical distinction 

 between these genera except in the petals, 

 which are entire in Arenaria and bifid in 

 Stellaria, and this difference becomes very 

 weak in certain species in which the petals 

 are merely emarginate, yet I_ doubt if any 

 systematist at present wishes to unite these 

 two large and traditionally maintained 

 genera. 



It is thus clear that by common consent 

 we are willing to preserve for the sake of 

 practical convenience certain familiar gen- 

 era, even though precise technical grounds 

 are lacking and a few intermediates occur. 

 In other words, there is an historic element 

 which will obtrude itself into our classifica- 

 tion. Its influence is obviously opposed to 

 a truly natural and symmetrical system; 

 nevertheless, much as it may be deplored 

 on theoretical grounds, it is ah undeniable 

 fact, and as such demands careful scrutiny 

 and consideration. 



Perhaps no large family shows the im- 

 portance of the historic element in generic 

 classification better than the Compositaj. 

 The lack of sharp boundaries between such 

 genera as Aster, Erigeron and Conyza, 

 Solidago, Aplopappus and Bigelovia, is 

 patent to all whose botanical experience 

 extends beyond the limits of some local 

 flora, yet so far as I am aware no botanist 

 has been willing to accept a recent proposi- 



tion to merge several of these genera into 

 one. The genera have through long recog- 

 nition become a practical aid in classifica- 

 tion. They are groups into which at least 

 a very high percentage of the component 

 species may be naturally associated, those 

 of intermediate character being, notwith- 

 standing their undeniable existence, rela- 

 tively little in evidence. 



But it may be asked whether the recogni- 

 tion of such somewhat ill-defined genera is 

 not quite at variance with our fine theories 

 that genera should be groups capable of 

 clear and mutually exclusive definition, 

 archipelagoes separated by deep and safely 

 navigable channels. Certainly this is quite 

 true, but I must beg you to notice that I 

 am in no wise urging any one to recognize 

 weak or poorly limited genera ; I am merely 

 stating the undeniable fact that certain 

 groups of this sort are now so widely recog- 

 nized as genera that we have no choice in 

 the matter. Aster, Erigeron, Conyza and 

 the like have become established facts in 

 classification. They fail to reach our 

 ideals, but this does not of necessity mean 

 that we must let down our standards else- 

 where. Attention may be called to the fact 

 that when they were first proposed these 

 genera were, so far as they were then 

 known, quite as distinct as could be desired ; 

 their present vague and merging state be- 

 ing due chiefly to intermediates discovered 

 long after the foiinding of the genera. 

 With the increase of knoAvledge these gen- 

 era which once seemed distinct have grown 

 together, the process has been a gradual 

 one and at no time has it seemed desirable 

 to abandon the idea of these genera, al- 

 though it has become increasingly evident 

 that it is to a great extent artificial. This 

 is one way in which the historic element 

 has entered into our classification. We may 

 readily admit that the aiTthors establishing 

 these old and now merging genera were in 

 most instances justified in doing so, because 



