238 W. L. DAVIDSON :

treat, or they might be grouped according to the language in which they are written, or they might be grouped according to the names of the authors alphabetically arranged. Each of these systems might plead a certain value, for even the last of them might conceivably bring out curious and practical statistical results. But we should hardly regard as legitimate any arrangement that proceeded on the mere colour of the bindings, or the number of letters in the authors' names, or the year in which the treatises were published, or the number of pages or of sheets that they contain. The very idea of a. Library (as distinct from a mere place for storing books) excludes such arrangements and brands them as ridiculous or capricious. In like manner, we may arrange the higher animals according to their nervous system or according to their intelligence ; but if we selected such an attribute as hairiness as the basis of our classification, we should lay ourselves open to the charge of arbitrariness or frivolity. From this it will be seen what an arbitrary or frivolous classification really means. The arbitrary and the frivolous, include not merely the fanciful and capricious, but also the accidental in all its forms, more particularly as the incon- stant. It is accidental to a book what the colour of its bind- ing or the number of its pages is ; and hairiness is a variable attribute among animals, differing even among individuals of the same species to almost any extent. No merely indi- vidual trait, no variable feature, no simple accident, can afford a rational basis of classification ; and all groupings that proceed upon one or other of these must be pronounced trifling and unscientific. Now, it is exactly from being based upon a trivial principle that many classifications, which from a purely formal point of view would be otherwise unimpeachable, are unsparingly condemned by the scientific classifier. It must be carefully noted that pure logical Division and Classification are not, in their whole length and breadth, coextensive. On the con- trary, it is sufficient for a logical Division, (1) that it be exhaustive, (2) that the parts be severally less than the thing divided, and (3) that the principle of Division be such as to secure that the parts be mutually exclusive. But what the character of the principle of Division itself is, beyond this fact of mutual exclusion, does not come within the ken of the formal logician. So that, when the formal logician adds to the three foregoing rules this fourth, viz., (4) that the prin- ciple of Division be important and essential, he does so by a sound enough instinct, but quite inconsistently with his own conception of the nature and scope of Logic.