in Natural History. 489 



Division and separation is the end of the artificial system ; — 

 to establish agreements is the end of the natural. In one case 

 we reason à priori ; in the other â posteriori. The one is a 

 descending, the other an ascending series. Linnanis under- 

 stood this distinction when he remarked, " Ordines naturales 

 valent cle naturâ plantarum ; artificiales in diagnosi plantarum." 

 — ■" Cavendo in imitando naturam filum Ariadneum amittamus." 

 Nevertheless it has appeared to me that many modern natu- 

 ralists have not adopted these truths ; and that it is the prevalent 

 error of the day to attempt to generalize where they ought to 

 analyse ; while their arrangements, called natural, are almost 

 all of them framed with a view to distinguish. Let me not be 

 supposed by these remarks to wish to exclude from the natural 

 system every attempt at diagnosis; for it is obvious, that as the 

 business of the naturalist is to study all the characters, he can 

 no more neglect ditferences than he can agreements. T only 

 wish to point out the two dissimilar objects we have in view, 

 that they may not be confounded. 



M. Decandolle, for instance, whose labours as a systematist 

 are invaluable, seems to overlook this distinction. In his " Regni 

 Vegetabilis Systeraa Naturale," he starts from things the least 

 known, to reason on things best known. He begins his compre- 

 hensive work with a predicate of the stars ; and, proceeding 

 downwards to minerals, comes to plants. Here he employs a 

 series of terms expressive of a natural gradation from the highest 

 to the lowest group, attempting fresh combinations at every stage, 

 and making a place for every thing. Thus he has class, sub-class, 

 cohort, order, tribe, gejius, section, species. 'The extraordinary 

 number of these combinations diminishes their value as a work of 

 natural arrangement. It is a difficulty of sufficient amount to 

 establish a few well marked ; and when they are so multiplied, it 

 may be suspected that many of them are arbitrary and artificial. 



3 u 2 This 



