information on the fact; the definitive conclusion depends
on the relation between the statements. We have, then, to study
separately the different cases which may occur.

III. Most frequently, except in contemporary history, the documents only
supply a single statement on a given fact. In such a case all the other
sciences follow an invariable rule: an isolated observation is not
admitted into science; it is quoted (with the observer's name), but no
conclusions are drawn from it. Historians have no avowable motive for
proceeding otherwise. When a fact is supported by no more than the
statement of a single man, however honest he may be, historians ought
not to assert it, but to do as men of science do--give the reference
(Thucydides states, Caesar says that ...); this is all they have a right
to affirm. In reality they all retain the habit of stating facts, as was
done in the middle ages, on the _authority_ of Thucydides or of Caesar;
many are simple enough to do so in express terms. Thus, allowing
themselves to be guided by natural credulity, unchecked by science,
historians end by admitting, on the insufficient presumption afforded by
a unique document, any statement which does not happen to be
contradicted by another document. Hence the absurd consequence that
history is more positive, and seems better established in regard to
those little known periods which are represented by a single writer than
in regard to facts known from thousands of documents which contradict
each other. The wars of the Medes known to Herodotus alone, the
adventures of Fredegonda related by none but Gregory of Tours, are less
subject to discussion than the events of the French Revolution, which
have been described by hundreds of contemporaries. This is a
discreditable state of things which cannot be ended except by a
revolution in the minds of historians.

IV. When we have several statements relating to the same fact, they may
contradict each other or they may agree. In order to be certain that
they really do contradict each other, we have to make sure that they do
actually relate to the same fact. Two apparently contradictory
statements may be merely parallel; they may not relate exactly to the
same moment, the same place, the same persons, the same episodes of an
event, and they may be both correct.[171] We must not, however, infer
that they confirm each other; each comes under the category of unique
statements.

If the contradiction