The work of abstraction by which, treating the facts directly known as
so much material for framing explanations, we pass from these actual
facts to the general laws which explain them, falls into four stages,
and at each stage, according to Bergson, as we go further and further
from the original fact directly known, the two vices of the
intellectual method, limitation and distortion of the actual fact,
become more and more apparent.

Starting from the fact directly known, the first thing, as we have
seen, is to learn to distinguish common qualities which it shares in
common with some, but not all, other facts; the next thing is to
classify it by fitting it into the further groups to which these
various qualities entitle it to belong. The moment a quality has been
distinguished in a fact that fact has been fitted into a class, the
class which consists of all the facts in which that quality can be
distinguished. Thus, in our original illustration, when you first
distinguished warmth, etc., you were beginning to fit your fact into
classes: when you perceived warmth you fitted it into the class of
warm objects, and it was the same with the other qualities of
roughness, size and smell. This fitting of facts into classes
according to the common qualities distinguished in them might be
called a preliminary classification, but we shall use the term
analysis for this preliminary grouping of facts according to their
qualities, keeping the term classification for the next step, which
you took when you realized "this is a dog," which consists in the
discovery not of mere disconnected qualities but of "real things."
Just as every quality, such as "warm" or "hairy" or "sweet" or "cold"
is a class of actual facts, so every "real thing" such as "a dog" or
"an ice cream" is a class of qualities. Thus a quality is once, and a
"real thing" is twice, removed from actual fact, and the more