e them, his text
only indicates how he wished to represent them, not how he really saw
them, still less how they really happened. What an author expresses is
not always what he believed, for he may have lied; what he believed is
not necessarily what happened, for he may have been mistaken. These
propositions are obvious. And yet a first and natural impulse leads us
to accept as true every statement contained in a document, which is
equivalent to assuming that no author ever lied or was deceived; and
this spontaneous credulity seems to possess a high degree of vitality,
for it persists in spite of the innumerable instances of error and
mendacity which daily experience brings before us.

Reflection has been forced on historians in the course of their work by
the circumstance of their finding documents which contradicted each
other; in such cases they have been obliged to doubt, and, after
examination, to admit the existence of error or mendacity; thus negative
criticism has appeared as a practical necessity for the purpose of
eliminating statements which are obviously false or erroneous. But the
instinct of confidence is so indestructible that it has hitherto
prevented even those professionally concerned from systematising the
internal criticism of statements in the same way as the external
criticism which deals with the origin of documents has been
systematised. Historians, in their works, and even theoretical writers
on historical method,[143] have been satisfied with common notions and
vague formulae in striking contrast with the precise terminology of the
critical investigation of sources. They are content to examine whether
the author was roughly _contemporary_ with the events, whether he was an
ocular _witness_, whether he was _sincere_ and _well-informed_, whether
he knew the truth and desired to tell it, or even--summing up the whole
question in a single formula--whether he was _trustworthy_.

This superficial criticism is certainly better than no criticism at all,
and has sufficed to give those who have applied it the consciousness of
incontestable superiority. But it is only a halfway-house between common
credulity and scientific method. Here, as in every science, the
starting-point must be methodical doubt.[144] All that has not been
proved must be temporarily regarded as doubtful; no proposition is to
be affirmed unless reasons can be adduced in favour of its truth.
Applied to the statements contained in docu