It should be remembered, however, that, unless a sound criterion be applied to the limitation of Families, they, like all other groups introduced into zooelogical systems, must forever remain arbitrary divisions, as they have been hitherto.  A retrospective glance at the progress of our science during the past century, in this connection, may perhaps help us to solve the difficulty.  Linnaeus, in his System of Nature, does not admit Families; he has only four kinds of groups,—­Classes, Orders, Genera, and Species.  It was among plants that naturalists first perceived those general traits of resemblance which exist everywhere among the members of natural families, and added this kind of group to the framework of their system.  In France, particularly, this method was pursued with success; and the improvements thus introduced by the French botanists were so great, and rendered their classification so superior to that of Linnaeus, that the botanical systems in which Families were introduced were called natural systems, in contradistinction especially to the botanical classification of Linnaeus, which was founded upon the organs of reproduction, and which received thenceforth the name of the sexual system of plants.  The same method so successfully used by botanists was soon introduced into Zooelogy by the French naturalists of the beginning of this century,—­Lamarck, Latreille, and Cuvier.  But, to this day, the limitation of Families among animals has not yet reached the precision which it has among plants, and I see no other reason for the difference than the absence of a leading principle to guide us in Zooelogy.

Families, as they exist in Nature, are based upon peculiarities of form as related to structure; but though a very large number of them have been named and recorded, very few are characterized with anything like scientific accuracy.  It has been a very simple matter to establish such groups according to the superficial method that has been pursued, for the fact that they are determined by external outline renders the recognition of them easy and in many instances almost instinctive; but it is very difficult to characterize them, or, in other words, to trace the connection between form and structure.  Indeed, many naturalists do not admit that Families are based upon form; and it was in trying to account for the facility with which they detect these groups, while they find it so difficult to characterize them, that I perceived that they are always associated with peculiarities of form.  Naturalists have established Families simply by bringing together a number of animals resembling each other more or less closely, and, taking usually the name of the Genus to which the best known among them belongs, they have given it a patronymic termination to designate the Family, and allowed the matter to rest there, sometimes without even attempting any description corresponding to those by which Genus and Species are commonly defined.