4. But observe, it is not enough in general that colour should be gradated by being made merely paler or darker at one place than another. Generally colour changes as it diminishes, and is not merely darker at one spot, but also purer at one spot than anywhere else. It does not in the least follow that the darkest spot should be the purest; still less so that the lightest should be the purest. Very often the two gradations more or less cross each other, one passing in one direction from paleness to darkness, another in another direction from purity to dullness, but there will almost always be both of them, however reconciled; and you must never be satisfied with a piece of colour until you have got both: that is to say, every piece of blue that you lay on must be quite blue only at some given spot, nor that a large spot; and must be gradated from that into less pure blue, - greyish blue, or greenish blue, or purplish blue, - over all the rest of the space it occupies. And this you must do in one of three ways: either, while the colour is wet, mix with it the colour which is to subdue it, adding gradually a little more and a little more; or else, when the colour is quite dry, strike a gradated touch of another colour over it, leaving only a point of the first tint visible; or else, lay the subduing tints on in small touches, as in the Exercise of tinting the chess-board. Of each of these methods I have something to tell you separately: but that is distinct from the subject of gradation, which I must not quit without once more pressing upon you the preeminent necessity of introducing it everywhere.

I have profound dislike of anything like habit of hand, and yet, in this one instance, I feel almost tempted to encourage you to get into a habit of never touching paper with colour, without securing a gradation. You will not, in Turner's largest oil pictures, perhaps six or seven feet long by four or five high, find one spot of colour as large as a grain of wheat ungradated: and you will find in practice, that brilliancy of hue, and vigour of light, and even the aspect of transparency in shade, are essentially dependent on this character alone; hardness, coldness, and opacity resulting far more from equality of colour than from nature of colour. Give me some mud off a city crossing, some ochre out of a gravel pit, a little whitening, and some coal-dust, and I will paint you a luminous picture, if you give me time to gradate my mud, and subdue my dust: but though you had the red of the ruby,the blue of the gentian, snow for the light, and amber for the gold, you cannot paint a luminous picture, if you keep the masses of those colours unbroken in purity, and unvarying in depth.

5. Next, note the three processes by which gradation and other characters are to be obtained:

A. Mixing while the colour is wet.

You may be confused by my first telling you to lay on the hues in separate patches, and then telling you to mix hues together as you lay them on: but the separate masses are to be laid, when colours distinctly oppose each other at a given limit; the hues to be mixed, when they palpitate one through the other, or fade one into the other. It is better to err a little on the distinct side. Thus I told you to paint the dark and light sides of the birch trunk separately, though, in reality, the two tints change, as the trunk turns away from the light, gradually one into the other; and, after being laid separately on, will need some farther touching to harmonise them: but they do so in a very narrow space, marked distinctly all the way up the trunk; and it is easier and safer, therefore, to keep them separate at first. Whereas it often happens that the whole beauty of two colours will depend on the one being continued well through the other, and playing in the midst of it: blue and green often do so in water; blue and grey, or purple and scarlet, in sky: in hundreds of such instances the most beautiful and truthful results may be obtained by laying one colour into the other while wet; judging wisely how far it will spread, or blending it with the brush in somewhat thicker consistence of wet body-colour; only observe, never mix in this way two mixtures; let the colour you lay into the other be always a simple, not a compound tint.

B. Laying one colour over another.

If you lay on a solid touch of vermilion, and, after it is quite dry, strike a little very wet carmine quickly over it, you will obtain a much more brilliant red than by mixing the carmine and vermilion. Similarly, if you lay a dark colour first, and strike a little blue or white body-colour lightly over it, you will get a more beautiful grey than by mixing the colour and the blue or white. In very perfect painting, artifices of this kind are continually used; but I would not have you trust much to them: they are apt to make you think too much of quality of colour. I should like you to depend on little more than the dead colours, simply laid on, only observe always this, that the less colour you do the work with, the better it will always be1: so that if you have laid a red colour, and you want a purple one above, do not mix the purple on your palette and lay it on so thick as to overpower the red, but take a little thin blue from your palette, and lay it lightly over the red, so as to let the red be seen through, and thus produce the required purple; and if you want a green hue over a blue one, do not lay a quantity of green on the blue, but a little yellow, and so on, always bringing the under colour into service as tar as you possibly can.