406 X. GENERAL BEIiLAJlKS. 



sort of selection will best meet the object ; namely, that 

 of rendering the comparisons passably complete or full, 

 yet without a confusing enumeration of too many orders. 

 A. De CandoUe has avowedly restricted his ordinal com- 

 parisons between different countries to the absolute num- 

 bers and per centage proportions of orders which together 

 constitute one-half of the flowering plants (Geog. Bot. p. 

 1190). This is a restriction to about ten orders, more or 

 less, and to those which usuallj'^ present less differences 

 than many others. It is objectionable, because com- 

 paring only half of the flora, and leaving out of view the 

 widest dissimilarities. There appears to be no particu- 

 lar advantage in ascertaining simply the orders which 

 predominate in various countries. If we would know and 

 seek to account for the characteristic differences between 

 countries, by aid of ordinal comparisons, it seems a better 

 course to ascertain which of the orders present the widest 

 dissimilarities ; those only being left out of view, which 

 aj)proximate to genera by the paucity of their species, 

 and the pettiness of their distinctions. 



Now, in almost every country, a minority of the orders, 

 — usually about a fourth, — will be found to include more 

 than the average number of species. Such over-average 

 orders include together the great bulk of the flora ; espe- 

 cially so in small countries, or more properly in small 

 floras, where many orders are usually represented by 

 single or few species to each. In Britain, for instance, 

 one-fifth of the orders are represented by single species, 

 and as many more by two or three ; scarcely half of the 

 orders having more than four species in each. Or, by 

 another mode of statement, the first 48 orders include 

 1320 species ; while the remaining 50 orders include 

 only 105 species. And dividing the 98 orders into three 

 groups, — 33, 33, 33, — these groups of orders include 



