doubtless have to invent a new type of law: law in duration, dynamic
relation. For we can, and we must, conceive that there is an evolution of
natural laws; that these laws never define anything but a momentary state
of things; that they are in reality like streaks determined in the flux of
becoming by the meeting of contrary currents. "Laws," says Monsieur
Boutroux, "are the bed down which passes the torrent of facts; they have
dug it, though they follow it." Yet we see the common theories of
evolution appealing to the concepts of the present to describe the past,
forcing them back to prehistoric times, and beyond the reasoning of today,
placing at the beginning what is only conceivable in the mind of the
contemporary thinker; in a word, imagining the same laws as always existing
and always observed. This is the method which Mr Bergson so justly
criticises in Spencer: that of reconstructing evolution with fragments of
its product.

If we wish thoroughly to grasp the reality of things, we must think
otherwise. Neither of these ready-made concepts, mechanism and finality,
is in place, because both of them imply the same postulate, viz. that
"everything is given," either at the beginning or at the end, whilst
evolution is nothing if it is not, on the contrary, "that which gives."
Let us take care not to confound evolution and development. There is the
stumbling-block of the usual transformist theories, and Mr Bergson devotes
to it a closely argued and singularly penetrating criticism, by an example
which he analyses in detail. ("Creative Evolution", chapter i.) These
theories either do not explain the birth of variation, and limit themselves
to an attempt to make us understand how, once born, it becomes fixed, or
else through need of adaptation they look for a conception of its birth.
But in both cases they fail.

"The truth is that adaptation explains the windings of the movement of