If we begin with some large class, such as ‘Animal,’ and subdivide it deductively into Vertebrate and Invertebrate, yet the principle of division (namely, central structure) has first been reached by a comparison of examples and by generalisation; if, on the other hand, beginning with individuals, we group them inductively into classes, and these again into wider ones (as dogs, rats, horses, whales and monkeys into mammalia) we are guided both in special cases by hypotheses as to the best grounds of resemblance, and throughout by the general principle of classification—­to associate things that are alike and to separate things that are unlike.  This principle holds implicitly a place in classification similar to that of causation in explanation; both are principles of intelligence.  Here, then, as in proof, induction is implied in deduction, and deduction in induction.  Still, the two modes of procedure may be usefully distinguished:  in deduction, we proceed from the idea of a whole to its parts, from general to special; in induction, from special (or particular) to general, from parts to the idea of a whole.

Sec. 4.  The process of Deductive Classification, or Formal Division, may be represented thus: 

A
|
-------------------
|                 |
A B               A b
|                 |
----------        ----------
|        |        |        |
A B C    A B c    A b C    A b c

Given any class (A) to be divided: 

1.  Select one important character, attribute, or quality (B), not common to all the individuals comprehended in the class, as the basis of division (fundamentum divisionis).

2.  Proceed by Dichotomy; that is, cut the given class into two, one having the selected attribute (say, B), the other not having it (b).  This, like all formal processes, assumes the principles of Contradiction and Excluded Middle, that ‘No A is both B and not-B,’ and that ’Every A is either B or not-B’ (chap. vi.  Sec. 3); and if these principles are not true, or not applicable, the method fails.

When a class is thus subdivided, it may be called, in relation to its subclasses, a Genus; and in relation to it, the subclasses may be called Species:  thus—­genus A, species AB and Ab, etc.

3.  Proceed gradually in the order of the importance of characters; that is, having divided the given class, subdivide on the same principle the two classes thence arising; and so again and again, step by step, until all the characters are exhausted:  Divisio ne fiat per saltum.

Suppose we were to attempt an exhaustive classification of things by this method, we must begin with ‘All Things,’ and divide them (say) into phenomenal and not-phenomenal, and then subdivide phenomena, and so on, thus: 

All Things
|
------------------
|                |
Phenomenal    Not-phenomenal
|
--------------------
|                  |
Extended          Unextended
|      (e.g., Pleasure and Pain)