r, till the person is found who first launched the
statement on its career, and with regard to him the question must be
asked: Was he an accurate observer?

Logically such a search is not inconceivable; ancient collections of
Arab traditions give lists of their successive guarantors. But, in
practice, lack of documents nearly always prevents us from getting as
far as the observer of a fact; the observation remains anonymous. A
general question then presents itself: How are we to criticise an
anonymous statement? It is not only "anonymous documents" with which we
are concerned, where the composition as a whole is the work of an
unknown author; even when the author is known, this question arises with
respect to each statement of his drawn from an unknown source.

Criticism works by reproducing the conditions under which an author
wrote, and has hardly anything to take hold of where a statement is
anonymous. The only method left is to examine the general conditions of
the document. We may inquire whether there is any feature common to all
the statements of a document indicating that they all proceed from
persons having the same prejudices or passions: in this case the
tradition followed by the author is biassed; the tradition followed by
Herodotus has both an Athenian bias and a Delphic bias. In respect of
each fact derived from such a tradition we must ask whether it has not
been distorted by the interest, the vanity, or the prejudices of the
group concerned. We may even ignore the author, and ask whether there
was anything likely to make for or against correct observation, common
to all the men of the time and country in which the observation must
have been made: for example, what means of information, and what
prejudices, had the Greeks of Herodotus' time with respect to the
Scythians.

The most useful of all these general inquiries has reference to that
mode of transmitting anonymous statements which is called _tradition_.
No second-hand statement has any value except in so far as it reproduces
its source; every addition is an alteration, and ought to be eliminated.
Similarly, all the intermediary sources are valueless except as copies
of the original statement founded directly on observation. The critic
needs to know whether this transmission from hand to hand has preserved
or distorted the original statement; above all, whether the tradition
embodied in the document was _written_ or _oral_. Writing fixes a
stateme