on, who
said, _History cannot be written from manuscripts_, which is as much as
to say: "It is impossible for a man to write history from documents
which he is obliged to put for himself into a condition in which they
can be used."

Formerly the professions of "critical scholar" and "historian" were, in
fact, clearly distinguished. The "historians" cultivated the empty and
pompous species of literature which then was known as "history," without
considering themselves bound to keep in touch with the work of the
scholars. The latter, for their part, determined by their critical
researches the conditions under which history must be written, but were
at no pains to write it themselves. Content to collect, emend, and
classify historical documents, they took no interest in history, and
understood the past no better than did the mass of their contemporaries.
The scholars acted as though erudition were an end in itself, and the
historians as if they had been able to reconstruct vanished realities by
the mere force of reflection and ingenuity applied to the inferior
documents, which were common property. So complete a divorce between
erudition and history seems to-day almost inexplicable, and it was in
truth mischievous enough. We need not say that the present advocates of
the division of labour in history have nothing of the kind in view. It
is admittedly necessary that close relations should obtain between the
world of historians and that of critical scholars, for the work of the
latter has no reason for existence beyond its utility to the former. All
that is meant is, that certain analytical and all synthetic operations
are not necessarily better performed when they are performed by the same
person; that though the characters of historian and scholar may be
combined, there is nothing illegitimate in their separation; and that
perhaps this separation is desirable in theory, as, in practice, it is
often a necessity.

In practice, what happens is as follows. Whatever part of history a man
undertakes to study, there are only three possible cases. In the first
the sources have already been emended and classified; in the second the
preliminary work on the sources, which has been only partially done, or
not at all, offers no great difficulty; in the third the sources are in
a very bad state, and require a great deal of labour to fit them for
use. We may observe, in passing, that there is naturally no proportion
between the intrins