119 



quickly assign any given plant to the group where its description 

 is to be found, and thus in many cases facilitates the work of 

 identification, for which reason it is still utilised to some extent 

 in modern Floras in the compilation of keys to the larger 

 natural groups. 



Striking characters, however, are often inconstant and by 

 themselves are not always trustworthy guides in classification, 

 while, apart from the fact that reliance on a single character 

 is apt to lead us wrong, it must be remembered that, if we 

 depend on one or two characters only, we are quite unable to 

 classify any plant unless the individual in question exhibits 

 those characters at the particular moment of its life- history 

 when we happen to see it. If, for instance, we rely only on 

 the number of cotyledons for the definition of primary groups 

 and the cotyledons are not to be found on our plant, we are 

 at once brought to a standstill and are unable to proceed with 

 the identification. We are then driven to the conclusion that, 

 as a rule, in botanical classification, an aggregate of characters 

 is of more value than one or two striking points of difference, 

 both in indicating general resemblance, and therefore relation- 

 ship, and in the work of identification. At the present day, 

 botanists endeavour to make all permanent classification as 

 natural as possible throughout, and hence, before deciding to 

 which group any plant belongs, all characters are as far as 

 possible taken into consideration and we then decide which 

 group it, on the whole, appears to resemble most, the plants 

 in any one natural group resembling, in the sum of theiv 

 characters, each other more than any other plant. 



107. Although, as has been The Unit of 



pointed out above, it has been recognised from a remote 9 assifica ' 

 period that the offspring of any given plant closely resem- 

 bles its parents, yet close observation showed that, in 

 reality, no two individual plants were ever exactly alike. If 

 then we accept similarity of form and structure alone as 

 the basis of classification, we must adopt the individual plant 

 as the unit. Since, however, every individual plant has 

 only a limited period of existence and sooner or later dies, 

 such a procedure would lead to no practical results. With 

 the individual as the unit we should, for instance, require an 

 unlimited number of names for all the individuals which 

 now exist or which will arise in the future from existing forms 

 by reproduction ; we should only be able to catalogue and de- 

 scribe a very small number of all the individuals ever existing 

 on the earth at one and the same time ; while the descriptions, 



