288 FIRST PRINCIPLES OF NATURAL CLASSIFICATIOX. 



into, the two adjoining ones. It may, indeed, be pos- 

 sible to discover a circular group witliout such collateral 

 helps ; but the discovery is highly improbable, and it 

 may be laid down as a rule that his first arrangement 

 will be more or less natural in proportion as he is 

 acquainted with the objects immediately surrounding, 

 or connected to, those which he is investigating. This 

 plan, moreover, of making out the circular series of 

 contiguous groups, is absolutely necessary for testing 

 the contents of that circle more immediately under 

 investigation. 



(350.) II. The second test to which our supposed 

 circle must be brought, is that of analogy; in other 

 words, those relations which its contents bear to the 

 neighbouring circles, and to all others in its own class 

 or order. It is an easy matter to place a series of 

 animals in a circle, and call it a natural group, and to 

 . repeat the same operation with such others as come 

 near to the first; but to make the contents, or divisions, 

 of these circles tally with each other is a very different 

 matter, and imposes a check upon the fancy which will 

 dissipate many illusions. No circle whose contents 

 will not bear such a comparison can be natural. It 

 may, indeed, happen, that one or even two of its sub- 

 divisions are wanting, while in the group with which 

 it is compared they are jjresent; yet even under these 

 circumstances there will be so strong a resemblance 

 between the two, in all other parts, that we may begin 

 to hope our arrangement of both is correct. We should 

 not, however, rest content with one or two tests of this 

 sort^ but bring our group, thus far safe, to encounter 

 all the comparisons which we can institute. Should it 

 be, for instance, the genus Picas; after tracing its 

 subgenera, or divisions, in the two neighbouring genera, 

 we should compare it with the sub-families of its own 

 circle, and then with the families of the Scansores. If 

 our arrangement is natural, we shall find parallel rela- 

 tions of analogy will result from these and all other 

 comparisons we make, and thus proceeding to the 



