764 



These characters are something like those employed by the French 

 ornithologist, M. Vieillot, in defining his order Grallatores. "Feet 

 moderate or long, robust, or slender. Legs half bare ; toes slit or 

 webbed, sometimes margined, two before only — three before only — or 

 three before and one behind ; pollex raised from the ground, or rest- 

 ing upon it only by the tip, or reposing upon it in its whole length ; 

 claws of various forms, not retractile ; bill of various shapes." !! So 

 much for a definite arrangement in Ornithology ; this is also termed 

 a "natural" system. 



My object in bringing forward these instances is neither to under- 

 value the ability of the eminent authors of these systems, nor the uti- 

 lity of well-constructed natural systems in their proper place ; but 

 merely to show the very vague definitions which have crept into our 

 modern classification, and the strange passages sometimes written 

 when authors are determined to see nothing but some shadowy ab- 

 straction, which cannot be reduced to practice. 



I cannot understand what is sometimes meant by the term, a " na- 

 tural group." It is said to be a group whose individuals agree in the 

 greatest number of points ; and moreover, botanists will tell us that 

 there is always a general resemblance among the members of a natu- 

 ral group (as indeed there must be if they are alike in the majority of 

 their characteristics) ; now Myosurus has but little resemblance to 

 Ranunculus or Aconitum, and yet they all belong to Ranunculaceae : 

 nor is Adoxa much like the ivy, yet both are classed in Araliaceae. 

 And if we examine the characters in detail, we shall find them as dis- 

 similar as the general appearance ; in fact, "natural" resemblances 

 have often no existence but in fancy. In such divisions as Compo- 

 sitee, CrucifersB, Graminese, MalvaceEe, Leguminosse, Labiatae, &c., 

 the species of which are very nearly allied, the general resemblance is 

 doubtless striking ; but let us not forget that these most generally 

 belong to Linnsean classes, only differing in name from those of the 

 natural scheme. 



Notwithstanding Decandolle's advocacy of the Linnaean axiom — 

 ".Genus dabit characterem, sed non character genus," — it is chiefly 

 his losing sight of this principle, which, though it may sometimes be 

 canied too far, is generally an admirable rule, that has occasioned 

 the introduction into his system of so many vague and useless cha- 

 racters. What Linnseus applies to genera, is equally applicable to 

 natural orders ; but instead of endeavouring to find a common cha- 

 racter or characters for the group to be established, it would almost 

 sucm, in some instances, as if a description were written, a number 



