soever.

Sec. 3. How used in composition.

But over these images, vivid and distinct as nature herself, he has a
command which over nature he has not. He can summon any that he chooses,
and if, therefore, any group of them which he received from nature be
not altogether to his mind, he is at liberty to remove some of the
component images, add others foreign, and re-arrange the whole.

Let us suppose, for instance, that he has perfect knowledge of the forms
of the Aiguilles Verte and Argentiere, and of the great glacier between
them at the upper extremity of the valley of Chamonix. The forms of the
mountains please him, but the presence of the glacier suits not his
purpose. He removes the glacier, sets the mountains farther apart, and
introduces between them part of the valley of the Rhone.

This is composition, and is what Dugald Stewart mistook for imagination,
in the kingdom of which noble faculty it has no part nor lot.

Sec. 4. Characteristics of composition.

The essential characters of composition, properly so called, are these.
The mind which desires the new feature summons up before it those images
which it supposes to be of the kind wanted, of these it takes the one
which it supposes to be fittest, and tries it: if it will not answer, it
tries another, until it has obtained such an association as pleases it.

In this operation, if it be of little sensibility, it regards only the
absolute beauty or value of the images brought before it; and takes that
or those which it thinks fairest or most interesting, without any regard
to their sympathy with those for whose company they are destined. Of
this kind is all vulgar composition; the "Mulino" of Claude, described
in the preface to the first part, being a characteristic example.

If the mind be of higher feeling, it will look to the sympathy or
contrast of the features, to their likeness or dissimilarity; it will
take, as it thinks best, features resembling or discordant, and if when
it has put them together, it be not satisfied, it will repeat the
process on the features themselves, cutting away one part and putting in
another, so working more and more delicately down to the lowest details,
until by dint of experiment, of repeated trials and shiftings, and
constant reference to principles, (as that two lines must not mimic one
another, that one mass must not be equal to another,) etc., it has
morticed together a satisfactory result.

Sec. 5. What pow