e liquid definitely blue: fill a large white
basin with the solution, and put anything you like to float on it,
or lie in it; walnut shells, bits of wood, leaves of flowers, etc.
Then study the effects of the reflections, and of the stems of the
flowers or submerged portions of the floating objects, as they
appear through the blue liquid; noting especially how, as you lower
your head and look along the surface, you see the reflections
clearly; and how, as you raise your head, you lose the reflections,
and see the submerged stems clearly.

[40] Respecting Architectural Drawing, see the notice of the works
of Prout in the Appendix.

LETTER III.

ON COLOR AND COMPOSITION.

152. MY DEAR READER,--If you have been obedient, and have hitherto done
all that I have told you, I trust it has not been without much subdued
remonstrance, and some serious vexation. For I should be sorry if, when
you were led by the course of your study to observe closely such things
as are beautiful in color, you had not longed to paint them, and felt
considerable difficulty in complying with your restriction to the use of
black, or blue, or gray. You _ought_ to love color, and to think nothing
quite beautiful or perfect without it; and if you really do love it, for
its own sake, and are not merely desirous to color because you think
painting a finer thing than drawing, there is some chance you may color
well. Nevertheless, you need not hope ever to produce anything more than
pleasant helps to memory, or useful and suggestive sketches in color,
unless you mean to be wholly an artist. You may, in the time which other
vocations leave at your disposal, produce finished, beautiful, and
masterly drawings in light and shade. But to color well, requires your
life. It cannot be done cheaper. The difficulty of doing right is
increased--not twofold nor threefold, but a thousandfold, and more--by
the addition of color to your work. For the chances are more than a
thousand to one against your being right both in form and color with a
given touch: it is difficult enough to be right in form, if you attend
to that only; but when you have to attend, at the same moment, to a much
more subtle thing than the form, the difficulty is strangely
increased,--and multiplied almost to infinity by this great fact, that,
while form is absolute, so that you can say at the moment you draw any
line that it is either right or wrong,