distinguish one group from another, For the purpose of classification
into "things" the quality can stand for the whole fact: thus, as
Bergson points out, we begin to lose contact with the whole fact
originally known, since all the rest of it except the respects in
which it can be analysed will henceforth tend to be ignored.

The third stage in explaining facts in terms of general laws is called
induction and consists in observing and formulating the relations of
"things." "Things" are related to each other through their qualities.
Qualities do not give us the whole fact, because, when we have
distinguished qualities, we are inclined to concentrate our attention
on the quality at the expense of the rest of the fact; nevertheless
while we attend to actual qualities we have not lost contact with fact
altogether. Induction, which consists in framing general laws of the
relations of "things," though it does not involve attention to the
whole fact, does at least demand attention to qualities, and so, while
we are occupied with induction, we do still keep touch with fact to
some extent.

Once the relations of qualities have been observed and formulated,
however, we need no longer attend to any part of the fact at all.
Instead of the actual qualities we now take symbols, words, for
example, or letters, or other signs, and with these symbols we make
for ourselves diagrams of the relations in which we have observed that
the qualities which they represent have stood to each other. Thus we
might use the words "lightning before thunder" or first an L and then
a T, to express the fact that in a storm we usually observe the
quality of flashing before the quality of rumbling. Such laws do not
actually reveal new facts to us, they can only tell us, provided we
actually know a fact belonging to a given class, to what other class