the subsequent imitative paroxysm
of crashing discords in chalk, lip-salve, and skim-milk, which has
lasted almost to this day.

At any rate, I throw out this hint for pupils and students, that if they
will get a set of glass samples and try combinations of colour in them,
they will have a bracing and guiding influence, the strength of which
they little dream of, regarding one of the hardest problems of their
art.

This for the student of painting in general: but for the glass-painter
it is absolutely essential--the central point, the breath-of-life of his
art.

To live in it daily and all day.

To be ever dealing with it thus.

To handle with the hands constantly.

To try this piece, and that piece, the little more and the little less.

This is the be-all and end-all, the beginning and the end of the whole
matter, and here therefore follow a few hints with regard to it.

And there is one rule of such dominating importance that all other hints
group themselves round it; and yet, strangely enough, I cannot remember
seeing it anywhere written down.

Take three tints of glass--a purple, let us say, a crimson, and a green.

Let it be supposed that, for some reason, you desire that this should
form a scheme of colour for a window, or part of a window, with, of
course, in addition, pure white, and probably some tints more neutral,
greenish-whites and olives or greys, for background.

You choose your purple (and, by-the-bye, almost the only way to get a
satisfactory one, except by a happy accident now and then, is to double
gold-pink with blue; this is the only way to get a purple that will
vibrate, palpitating against the eye like the petal of a pansy in the
sun). Well, you get your purple, and you get your green--not a
sage-green, or an "art-green," but a cold, sharp green, like a leaf of
parsley, an aquamarine, the tree in the "Eve" window at Fairford, grass
in an orchard about sunset, or a railway-signal lamp at night.

Your crimson like a peony, your white like white silk; and now you are
started.

You put slabs of these--equal-sized samples, we will suppose--side by
side, and see "if they will do."

And they don't "do" at all.

Take away the red.

The green and the purple do well enough, and the white.

But you _want_ the red, you say.

Well, _put back a tenth part of it_.

And how now?

Add a still smaller bit of pale pink.

And how now?

Do you see what it all means? It means the rule we s