had when setting out to fight
against the Visigoths. How are we to make our imagination of facts of
this kind harmonise with the reality?

Practically, what happens is as follows. Immediately on the reading of a
sentence in a document an image is formed in our minds by a spontaneous
operation beyond our control. This image is based on a superficial
analogy, and is, as a rule, grossly inaccurate. Any one who searches his
memory may recall the absurd manner in which he first represented to
himself the persons and scenes of the past. It is the task of history to
rectify these images gradually, by eliminating the false elements one by
one, and replacing them by true ones. We have seen red-haired people,
bucklers, and Frankish battle-axes (or at least drawings of these
objects); we bring these elements together, in order to correct our
first mental image of the Frankish warriors. The historical image thus
ends by becoming a combination of features borrowed from different
experiences.

It is not enough to represent to oneself isolated persons, objects, and
actions. Men and their actions form part of a whole, of a society and of
a process of evolution. It is, therefore, further necessary to represent
to oneself the relations between different men and different actions
(nations, governments, laws, wars).

But in order to imagine relations it is necessary to have a conception
of collectivities or wholes, and the documents only give isolated
elements. Here again the historian is obliged to use a subjective
method. He imagines a society or a process of evolution, and in this
imaginary framework he disposes the elements furnished by the documents.
Thus, whereas biological classification is guided by the objective
observation of physical units, historical classification can only be
effected upon subjective units existing in the imagination.

The realities of the past are things which we do not observe, and which
we can only know in virtue of their resemblance to the realities of the
present. In order to realise the conditions under which past events
happened, we must observe the humanity of to-day, and look for the
conditions under which analogous events happen now. History thus becomes
an application of the descriptive sciences which deal with humanity,
descriptive psychology, sociology or social science; but all these
sciences are still but imperfectly established, and their defects retard
the establishment of a science of hi