328 PRACTICAL AND SCIENTIFIC ZOOLOGY. 



other hand, so numberless are the forms of nature, that 

 false circles can be made, and are frequently made,, by 

 putting in, to fill up our gaps, animals which have no 

 real connection with that circle which we wish to ren~ 

 der perfect. Hence, although we must first look to the 

 circularity of a group as a primary requisite, still the 

 accuracy of this circle must be proved by other tests, 

 which will be shortly explained. 



(400.) The second property possessed by natural 

 groups regards those only which we call aberrant, and 

 consists in the three aberrant groups or divisions of a 

 circle being united among themselves into one circle, 

 independent of their union also with the two typical 

 groups. This theory, although it virtually makes the 

 primary division of every circle to be three, does not, 

 in fact, affect the accuracy of a group which is first 

 divided into five, any more than this, that it shows 

 these aberrant divisions to have other properties than 

 were formally suspected ; so that, besides being united 

 to the typical groups, they also blend in a circle of their 

 own, as if they were independent of the two others. 



(401.) As we have hitherto looked to the vertebrated 

 animals as furnishing one of the most familiar illustra- 

 tions of natural arrangement, we will again use them to 

 exemplify the union of which we are now speaking. 

 Quadrupeds and birds, then, are the two typical groups 

 of vertebrated animals ; while reptiles, amphibia, and 

 fishes are the three aberrant. Now, if these latter are 

 found, upon investigation, to form a circle by themselves, 

 it naturally follows that the primary circles in every 

 group are three, and not five ; the three aberrant divi- 

 sions being merged into one. This union, however, 

 cannot always be traced, from the causes elsewhere 

 assigned ; and therefore, in dubious cases, it is more ad- 

 visable to adhere to the usual method of distinguishing 

 each of the aberrant groups separately by themselves. It 

 follows, nevertheless, that, wherever it can be demon- 

 strated, we must consider that the circle is first divided 

 into three others, each of which is again resolved into, 



