Notebook, 1993-


Thompson, Daniel V., Jr., Research and Technical Adviser, The Courtauld Institute of Art, University of London. The Practice of Tempera Painting. New Haven: Yale University Press. 1936. Fourth Printing, 1946.

The Practice of
Tempera Painting - Painting (cont.)

Systematic Value Mixing Essential
It is likely to be regarded as an intrusion upon the sacrosanct privilege of the individual to insist that there are right ways and wrong ways of mixing intermediate values. Nevertheless, if you want a series of three values, 1-3-5, mix l light, and 5 dark, and compound the intermediate 3 by mixing some of 1 with some of 5. If you want a series of five, create 2 by mixing some of 1 with some of 3, and 4 by mixing some of 3 with some of 5; and use the same principle if you want a still longer series. This is not an arbitrary whim: there is a very practical reason for doing it this way, and not, as painters are used to doing, by eye. If the lightest value, 1, is white, and the darkest, 5, is black the reason does not appear; but if 1 is white and 5 is pale yellow, it will immediately be found, on experiment, that a graded series cannot be mixed by eye alone. Even in less extreme cases than this, it will be found difficult or impossible to distinguish between close adjacent values as they stand in the color cups. But mixtures which look identical in the color cups tell as very different when they are painted out. It is possible, and sometimes extremely useful, in tempera to model very fully and completely within a very narrow value range. It is quite possible to model with a series of three or five or seven value steps between white and pale yellow or pale gray, and to have each value count in the modeling. Some of the most precious effects of tempera painting depend on this power of compressing [p. 111] the value scale without reducing the completeness of the rendering. The smoothness of the surface and the clarity of the painting make every variation of tone effective; and it is quite practicable in tempera to model fully between value limits which in other media would preclude the use of any modeling at all. It is, however, necessary to mix the members of the value series systematically, and to keep the cups in order during the painting process. [pp. 110-110]

Variations of Color System
Mixtures of a single color with white have been described as the simplest basis of manipulation, and so they are. [The single color may, of course, be a simple pigment or a mixture of pigments.] One is often content to model an area from vermilion in the shadows to white in the lights, or from opaque oxide of chromium to white. Modeling of this sort is perfectly descriptive, though arbitrary. Sometimes, however, one wants an entirely different sort of effect. It is desirable from the point of view of mechanical handling and also of orderliness in the painting to achieve one's effects by some systematic means; and when the basic system, pure color to white, will not serve, to substitute some other equally systematic method. It is inconvenient in practice to try to mix each value by eye to a separate color, and the result of an attempt to do so is not likely to be happy.

Two modifications of the pure-color-to-white system are adequate to many purposes. First, the substitution of some other color for white. instead of modeling vermilion up to white, it may be modeled up to yellow, or a mixture of white and yellow, for example; and green in the same way. Or green may be modeled up to light blue, or to a lighter green of different composition. This system requires only the substitution of the new light for white in the other method, but it produces a totally different series of colors, and introduces a marked color change accompanying each change of value. The second modification is to put the pure color in the middle value instead of in the dark. Vermilion in the reflected lights, for example, may be mixed with white in the half tones and modeled [p. 111] up to white in the lights; while the shadows are made with madder red or alizarin crimson. Oxide of chromium may be modeled up to cadmium yellow in the half tones and lights, and given viridian or cobalt blue for its shadow. Cobalt blue may be modeled up to white [or pink, or yellow, or anything you please], and down to purple, or black, or dark green. The intermediate values, if any are required, between the pure color in the middle and the other colors at the top and bottom of the scale, are made as before by mixing.

These simple, systematic bases for color and value series permit a wide variety of color effects without complicating the manipulation of the medium. Others may readily be devised. Adopting a systematic modeling scheme imposes no greater hardship on a painter than the conventional divisions of the musical scale impose upon the musician. If a composer cannot make shift to write music within the limits of the keyboard, he need not look for a performance of his works on the piano. If a painter feels the need of more liberty in the matter of color changes than is compatible with any simple system of pigment mixing, he will not find satisfaction in tempera painting. There are many worthy points of view about painting, and tempera is only one of many methods. Good tempera painting depends on a painter's willingness to accept certain limitations in return for certain powers. Other media offer other powers--and impose other limitations. [pp. 111-112]

Treatment of Surface Effects
It is perhaps hardly necessary to say that surface effects, such as the pattern in a drapery, should be painted in after the main forms are modeled; but the same principle can often be extended widely. Paint the neck and add the necklace; paint the waist and add the belt; paint the head and then the veil which hides it, the sky, and then the branch across it, the floor, and then the tiles, the bowl, and then the pattern on it. Even such an extreme subject as a Scotch plaid will yield to treatment of this sort! Necessary accidents of color, as in reflections, may be inserted at the end. The painter of naturalistic effects will not wish to work this way; but it is a good way, all the same. [p. 112]

Painting Flesh
I have and treasure some little nineteenth-century booklets on painting which describe exactly how to paint landscapes and portraits. They tell me just what colors to mix together for every purpose; their authors almost hold my brush. They could hardly be worse than they are, or more dogmatic. Yet I feel that I may seem to be running into the same sins. I have described the painting of a sky, almost stroke by stroke, and am about to describe the painting of a head. Let it be understood that these notes are not meant as formulas but as illustrations. I have no special fondness for paintings of skies graded from dark blue to white; I do not want them painted to my order. In my little 1860 booklets which tell how to paint foliage and hair and satin dresses and running water and sunset effects there may be found some sage and useful counsel on the handling of oil paint. Perhaps there is no other way to teach the nature of a medium than in terms of practice, at the risk of seeming to expound a dogma. I am going to describe the painting of a head in the Giottesque manner; but I do not want to see a modern painter turn Giottesque. There is something to be learned from this method of Cennino's, in painting flesh, something that a modern painter may absorb and turn to good account; but it calls for metamorphosis.

If you have modeled a head carefully in ink on the white surface of the gesso, and you pass over it, as Cennino directs, two coats of terre-verte and white, well tempered, you will have the effect of an underpainting in green and greenish gray. Terre-verte is a particularly transparent, fat color. Some samples are quite warm, almost yellow, and other cold and blue. Upon the exact shade of the terre-verte depends the exact color of the pearly, opalescent, greenish gray which appears in this underpainting wherever the paint lies over the darker drawing; but whatever the quality of the sample, it will yield some sort of pearly tone all over the half tone and shadow areas. Curiously enough, this tone does not depend on the color underneath: that may be ink or paint, brown, or black, or red, if you please [though there is no need to use anything but india ink]. The effect of the green over it will be almost exactly the same in any [p. 113] case; for this opalescence depends only on the darkness of the color underneath.

If you mix a little brown, say yellow ocher with a touch of black and earth red and little white, temper it, and thin it out well with water, you can shade and model this pearly half tone down and around into shadow, making it warmer, darker, and more transparent at the same time. The play of dark, transparent warm over lighter, opalescent cool will give you perfect control over the modeling of the shadow both in form and color, and in transparency as well. You will want to take pains not to lose the pearly, silvery quality of the half tone by carrying the brown shading [which Cennino calls "verdaccio"] too far. Where you place the shading depends, of course, on how you are handling the light and shade; but whatever your system in that may be, you will have some place for every degree from the untouched scumble of green to the full strength of your verdaccio.

Next you may want to mix a color for the redness of the cheeks, and lay it in place, carrying it thinly as far into the shadow as you wish, and establishing it strongly in the lights. Then mix a general flesh tone, or two, or three, if the case seems to you to require it, and work the lights up to the edge of the half tones or shadows. Mix as many lighter flesh tones as you need, and build up the lights with them, reinforcing variations of local color in the lips and cheeks and chin and forehead as you go. At the end, you may have some crisping up to do with verdaccio, or other dark color, and light colors, or pure white. You will follow the universal principle of developing the form gradually, keeping it open and plastic, under your control, until you are ready to make it definite and fixed by a few decisive strokes. [pp. 113-114]

Painting Hair
In painting flesh, do not try to stop short at the hair. Paint the flesh well over the outline of the hair, and then paint the hair back in again. The two coats of terre-verte may as well be laid over the hair as not. When the head is painted, restore the drawing of the [p. 114] hair with any dark color you please, verdaccio or black, it makes no difference what , and lay one or two coats of whatever local color you wish over it. The dark drawing will show through a little, and cause the ocher or red or brown or gray of the hair to exhibit the indispensable opalescent half tone upon which you work your shadows and reflected lights. Paint in the direct lights opaquely with suitable color; and if you like, finish with some drawing strokes of fairly thick paint. [pp. 114-115]

Variations of Method
This is not to be regarded as a formula, but as a system subject to much variation. You may not wish to underpaint with terre-verte. You may not like the half tone that green gives. If it is too green for your taste, use brown or yellow or red or blue or anything you please. [Blue is a superb preparation for the painting of black flesh.] If it is not green enough to suit you, use viridian! The whole point is to create some relatively cool half tone, midway between transparent and opaque, and this can be done with any color in your palette. You are at liberty, within the system, to make your colors anything you please, to apply them delicately or boldly, in an abstract scheme of modeling or in the most scientific order of light and shade. The adoption of the basic system simply gives you control over warm and cool and transparent and opaque with a minimum of labor and a maximum of flexibility,.

This is the whole purpose of the manipulations described in this chapter: to achieve control of the effects of painting through the systematic use of the materials. There are other systems which may be used. You may lay in all the areas of the painting in flat tones of middle value, and model them up and down with lighter and darker values. That is a good method, especially for small sized works. In place of the monochrome ink drawing on the panel, you may lay in your shadows first with dark local color, scumble them over with a lighter value, and restore the transparency with a glaze ad lib.; and that is not a bad method at all. Every painter will find his own level, work out his own applications, or reject the medium [p. 115] entirely. It is not suitable to everyone, or to all purposes. When it is suitable, however, it offers an economical means of producing work of exceptional power and beauty. [pp. 115-116]

Final Embellishments
It remains only to speak of mordant gilding, and the processes of painting and embellishing most useful to the panel painter will all have been discussed. We have seen how broad grounds of metal may be laid and burnished; how patterns of gold and color may be formed by painting over the burnished gold, scraping away the color in a pattern, and graining the exposed gold. We have considered the use of powdered gold as a pigment, for modeling on other colors. Now let us look at a further use of gold, gold laid in the form of leaf in a determinate pattern on the top of color. This type of gilding, called "mordant" gilding because it is mordanted in place, differs strongly from burnished gold and from shell gold in effect. It may be used for large areas, especially as a ground for glazings of color, usually in oil; but its characteristic application is in rather fine lines or dots or small shapes on a colored ground. The principle is to work the design out in a sticky material, the mordant, and then to cover it with gold leaf, brushing away the gold leaf from the parts not covered by the mordant. This method gives the effect of an incrustation of gold upon the painted surface. [p. 116]

Water Mordants
There are two types of mordant for this purpose, oil mordants and water mordants. Oil mordants are more durable, and on the whole more satisfactory than water mordants; but they are more troublesome to prepare in the first instance, and a little more exacting in use. A very good water mordant may be made by pounding cloves of garlic in a mortar and squeezing out the juice, through a linen cloth, adding a little powdered gum ammoniac [which may be left in it overnight to soften and dissolve], and grinding to a fairly stiff paste with white lead and a little bole. The gum ammoniac is not absolutely necessary, but is an improvement. [p. 117]

Oil Mordants
No commercial gold size is altogether satisfactory as a mordant for use on tempera paint. Japan gold size is too thin when used alone. It tends to run and spread, making untidy lines. It dries too quickly and unevenly on the tempera surface. The fat-oil gold sizes are better, but they generally dry too slowly for convenience [often requiring a week or more to develop the right tack], and contain too little pigment to make a good line. Pigments ground in Japan, such as chrome yellow, have body enough, but tend like Japan alone to paint out too thinly, or else to make a crude stroke. Chrome yellow in Japan mixed with a good fat-oil size and a drop or two of strong copal varnish gives a mixture which will flow out of a brush in a fine line and settle to a smooth enamel surface without spreading. One has to experiment with the mixtures until he is satisfied. Of the commercial preparations, those sold as "Quick oil gold size" are probably the most suitable. For very fine and delicate work, it is desirable to have a mordant which will "thread" from the brush, and yet be sticky enough to hold well on the painted surface. This is more difficult to achieve. It can be managed, however, by boiling such a mixture of pigment fat-oil, varnish, and Japan as has been described, cooling it, and boiling it again repeatedly until it is almost rubbery in consistency.

Another way is to boil a small amount of linseed oil [out-of-doors, or under a fume hood, for it makes a very bad smell], until it is about as thick as strained honey; grind it, as Cennino directs, with some white lead and, if you like, a little verdigris; add a little powdered sandarc resin; and boil the mixture together for a few minutes. The consistency may be varied by adding more oil or more resin, until it suits you. It is quite impossible to give any definite [p. 117] rule for mixture of his sort made in small quantity. One need hardly make up a mordant more than once in a lifetime; for an ounce or two, tightly stoppered, will last indefinitely, and a very little goes a very long way. [pp. 117-118]

Applying the Mordant
Whichever type of mordant one uses, the application is much the same. A brush is chosen in proportion with the scale of the ornament to be applied, and the mordant very carefully and neatly laid on. It is a good plan to brush a little glair thinly over the section of the painting which is to receive oil mordant gilding, and let the glair dry before the mordant is applied. This thin film of glair will keep the gold from sticking to the surface of the paint. With a water mordant this is not a sufficient safeguard, and if the surface of the painting seems at all sticky or greasy, it should be dusted lightly with powdered French chalk before the mordant is applied. If the mordant tends to soak into the surface, it will dry too quickly and unevenly, and the painting should be sized locally with one or two thin coats of one-to-sixteen gelatine solution, or with a thin solution of gum benzoin or shellac in alcohol, to render it less absorbent. [p. 118]

The water mordants dry promptly, and may be gilded soon after application. Indeed, they often dry so quickly that it is necessary to breathe lightly on them to restore the stickiness. This breathing should not be carried further than necessary, as it may make the mordant too wet. Oil mordants, on the other hand, require some time to dry. Japan alone may be too dry in half an hour, and a fat-oil gold size may be too soft at the end of a week. It is impossible to predict the amount of time required for drawing; but for a given mordant it will not vary much, and when you know your material you can tell whether to plan to gild in eight hours, twenty-four, forty-eight, or whatever it may be. It is most convenient to have a mordant dry in twelve to eighteen hours; for them what you lay one day will be just ready to be gilded the next morning. The gold should be laid as soon as the mordant is tacky. You should be able [p. 118] to run your finger up lightly along a line of the mordant without smearing it, yet it should be tacky in the thinnest parts when you touch it with slight pressure. Cut the gold on the cushion as required. Lay it along the mordant with the tip, overlapping the leaves always in the same direction. When all the mordant ornament is covered with gold, press the gold down with a camels' hair mop and then with a bit of cotton, lightly at first, then harder. Apply more gold if you find it necessary. Then take the mop or a fair-sized sable brush, and delicately skew away the loose gold, and work the skewings back and forth over the mordant gilding so that they will stick to any tiny places which have not taken the gold. It is often wise to gild while the mordant is a little fresh, and then to leave it for a day or two to dry and harden before dusting off the excess gold. To lay the gold on the mordant, cut the leaf as required, and lift it by the edge, as shown in the drawing on page 126, below [in the chapter on Varnishing and Framing]. [pp. 119-119]


[Thompson, Daniel V., Jr., Research and Technical Adviser, The Courtauld Institute of Art, University of London. The Practice of Tempera Painting. New Haven: Yale University Press. 1936. Fourth Printing, 1946.]



The contents of this site, including all images and text, are for personal, educational, non-commercial use only. The contents of this site may not be reproduced in any form without proper reference to Text, Author, Publisher, and Date of Publication [and page #s when suitable].