THE CLASSIFICATION OF PLANTS 2129 



size and composition but in their analysis it would be wise to use a ter- 

 minology which is free from confusion and to leave the term " species " 

 where we found it, in the museum. For this reason the deme of Gilmour 

 and Gregor has much to recommend it. The root " deme ", meaning a 

 population, is non-committal in itself and it can be suitably qualified by 

 prefixes as experimental study may justify. Thus, a gamodeme is a freely 

 interbreeding population, an ecodeme an ecologically localized population, a 

 gefwdetne a genetically differentiated population, a cytodeme a cytologically 

 differentiated population (cytotype) and an ecogenodeme would be the 

 equivalent of an ecotype. This kind of system is indefinitely flexible, it 

 makes no inroads upon classical taxonomy and it frees the mind of the 

 investigator to concentrate on the processes of speciation rather than on the 

 delimitation of categories. 



Turning now from the analysis of the species as the fundamental unit of 

 taxonomy, what of the relationships of the larger groups? Hayata, the 

 originator of the Dynamic System of Classification, pointed out in 1921 that 

 a natural system must be a network, not a phylogenetic tree, as the affinities 

 of groups depend upon the extent to which they share characters. From 

 this standpoint no groups can be regarded as having a permanently fixed 

 place in a system, since both the outline of the group and its position will 

 depend upon the number of characters taken into consideration. If atten- 

 tion be concentrated upon a few characters, classification will be relatively 

 easy, and this is, in fact, the way in which most of the systems referred to 

 later in this chapter have been built up. However, as we extend our scope 

 of consideration, so will our ideas of relative affinity change and the more 

 numerous the characters used, the more difficult will it become to make any 

 scheme, even in three dimensions, which will express all the relationships 

 involved. 



Hayata himself proceeded by grouping around each plant family all 

 the others which showed evidence of characters shared with it. Any one 

 familv may thus appear in several groups and the whole pattern is kaleido- 

 scopically changed as we turn from one set of characters to another. This 

 accords very well with the experience of practical systematists in attempting 

 to unravel the tangle of cross-relationships in any group of plants. No 

 fixed system, not even a three-dimensional lattice, can adequately express 

 them all; there is a hyper-dimensional quality about them on which we can 

 only speculate. We need a form of space in which an object can be in a 

 number of different places at the same time. There is, indeed, as Fries 

 said long ago: qiioddam siipernatiirale in systema naturae. 



All this may be true but it is certainly not practical systematics. We 

 return again to our dilemma and to the old saying, " il faut se borner". A 

 working classification must be to some extent artificial, an abstraction from 

 Nature, founded upon an arbitrary limitation of the criteria included. So 

 long as this is clearly understood and we are not thereby blinded to the 

 wider significance of relationships, the procedure is legitimate and indeed 

 inevitable. To keep our minds clear, however, we must recognize that in 



