100 The Scottish Naturalist. 



a generic character, and so forth, we should say a class-mark, a 

 generic mark, &:c.), and "character" would be applicable to any 

 one of the mark's constituents, and the sum-total of characters 

 would be the mark proper. 



But a point of more importance arises when we come to con- 

 sider what characters are worthy of being regarded as indispensable 

 to the mark. Are the necessary constituents simply those that 

 are possessed by every member of the group ? or shall we further 

 include those that are shared by the majority of members, though 

 not by all ? No doubt the distinction between universal and 

 general characters would be a very desirable one, if only it could 

 be kept up. But, unfortunately, it is not workable. In many 

 species and genera, indeed, it might be pretty rigorously applied 

 (and there it should be rigorously adopted), but it fails when we 

 reach the higher grades, classes, sub-classes, &c. Thus, for 

 instance, there is not a single character in the class Dicotyledons 

 that all Dicotyledons have in common. Even the fact of dicoty- 

 ledonism fails in the case of the Dodder, which is acotvledonous 

 (though not of course in the same way as the Acotyledonia or 

 Cryptogams), and in Abronia, which is monocotyledonous ; and 

 if we include Conifers and other Gymnosperms in this class, we 

 have a large group of /^^cotyledonous plants. But if dicoty- 

 ledonism fails, much more does every other character as yet 

 discovered, for there is no other fact of equal generality with it. 

 The mode of growth varies to an indefinite extent j so does the 

 venation ; and so does the floral symmetry. Universality then is 

 a thing not to be thought of, and we must rest content with bare 

 generality. Yet care should be taken that this generality is real, 

 and not merely apparent ; for a feature may be striking, and may 

 be shared in by a large number of plants, and nevertheless this 

 large number may be simply the minority. There is nothing but 

 a quantitative measure to go by, and a general character is the 

 same thing as a character possessed by a majority of members of 

 a group. 



Under these circumstances, various things become necessary. 

 It is indispensable (i) to exhaust the characters at every grade in 

 so far as this is possible ; (2) to indicate the exceptions, when the 

 characters are not absolutely universal, or, where it would be too 

 cumbrous to give a full enumeration, to supply good typical 

 instances; and (3) rigorously to exclude from the denning mark 

 of a grade irrelevant characters — i.e. (a) every particular that pro- 

 perly belongs to the grades subordinate as well as to the grades 



