22H 



THE AMERICAN NATURALIST [Vol. XL1II 



that these intergrades are what might be expected, and 

 they occur in the general region which, according to the 

 vascular anatomist, gave rise to the monocotyledonous 

 offshoot. It has always interested me to see how con- 

 vinced we become by our own definitions. We have legis- 

 lated that the last resort for distinguishing monocotyle- 

 dons and dicotyledons is the cotyledon character ; all other 

 characters have been found to be liable to exception. I 

 submit it to you whether any single character selected in 

 this way as final arbiter could not function equally well 

 as a character of last resort. This business of last-resort 

 characters is nothing less than harking back to an arti- 

 ficial system. It is hardly conceivable in these days that 

 such a character can really exist. It is the totality of 

 characters that must place an organism, a most difficult 

 test to apply, but none the less essential. A conspicuous 

 illustration of this situation is that of Selaginella. It is 

 assumed that the last-resort character of a seed-plant is 

 the seed ; and yet no definition of a seed can be constructed 

 that will exclude all species of Selaginella and include all 

 seed-plants. Then why is not Selaginella a seed-plant? 

 Simply because its other characters forbid such an asso- 

 ciation. There is no conceivable reason, therefore, why a 

 dicotyledon may not be monocotyledonous and still remain 

 a dicotyledon, or vice versa. The vascular anatomist tells 

 us that one of the surest marks of a monocotyledon is the 

 amphivasal bundle; and at the same time he points out 

 amphivasal bundles among dicotyledons. 



I am pressing this point perhaps unduly, but there is a 

 growing tendency that should be checked. This is to 

 transfer groups on a single character, or to propose 

 phylogenetic connections without weighing or waiting for 

 all the characters involved. It is easy to construct a satis- 

 factory scheme based upon one character ; it has thus far 

 proved impossible to construct a satisfactory scheme 

 based upon all the characters we happen to know. 



The spirit that animates modern morphology is no- 

 where more evident than in its effect upon teaching. 



