The image of a sky blue is formed entirely by the blue film, which is very weakly coloured in this part, thus reproducing tIie blue (mixed with white) of the subject. The image of a myrtle-green will be formed by the superposition of the yellow and of greenish-blue of full depth, and of pink of somewhat less intensity. The pure green resulting from the superposition of the yellow and greenish-blue will be greatly darkened by the pink which is its complemen tary, but of which the intensity is insufficient to effect complete extinction. The dull green of the subject will thus be formed.
870. Similarities of the Two Methods of Synthesis. The following experiment shows the close correlation between synthesis by addition of lights and subtractive synthesis by super position of pigments.
Three projection lanterns being adjusted as assumed in § 867 and fitted each with its coloured filter, let us place only one of the transparencies in its corresponding lantern. If, for instance, we place in the lantern projecting green light the transparency printed from the negative exposed through the green filter, it will he found that the image on the white screen appears a bright pink, which is a mixture of the blue and red lights freely passed by the lanterns in which there are no transparencies. In the same way, if only the transparency corresponding with the blue filter, or with the red filter, is placed in position, the image on a white screen will appear as yellow (mixture of green and red) or as greenish-blue (mixture of blue and green).
The introduction of one transparency in the beam of light corresponding with it stops the rays which the partly masked filter would have transmitted, and by the action of which the corresponding negative was made. Time colour which results from this subtraction is naturally the complementary of that of the selection filter.
Tile simultaneous introduction of two trans parencies, each in the beam corresponding with it, removes from the parts of the image common to both the rays which the two filters thus masked transmitted, thus leaving only the light transmitted by the third filter. This last colour is exactly that which would result from the superposition of the two pigmentary images, respectively complementary to the marked filters. If, for instance, the only filter the light from which is not thus intercepted by a trans parency is the green filter, the green which on the screen illuminates the parts common to the two projected images is the one that would be obtained by superposing a yellow monochrome (complementary to the blue filter) and a greenish blue monochrome (complementary to the i ed filter).
871. Impossibility of Trichromatic Reproduc tion of the Spectrum. The trichromatic pro cesses, while reproducing and modulating satis factorily pigmentary colours, at least when their bands of complete absorption and reflection are not very narrow,' fail to give even a grossly approximate rendering of the spectrum.
When an attempt is made to reproduce the spectrum, it is found that the latter is divided into five flat tints. The three regions, each
transmitted by only one selection filter, are represented by the colour of this filter. The two intermediate bands are rendered differently according as the spectral-transmission regions of the selection filters have, or have not, regions in common. If such regions overlap, the colour of the intermediate bands is that of the addition of the lights transmitted by the respective filters; if these regions have a gap between them, the intermediate bands are black.
By reason of its very principle, trichrome photography cannot reproduce a colour which is more elementary than those of the filters used for analysis. This impossibility is of minor account, since elementary colours do not exist in Nature. The rainbow and the colours due to interference and polarization are sufficiently far removed from elementary rays to permit of being reproduced satisfactorily.
872. Photographs in Two Complementary Colours. While it is obviously impossible, by employing two colours only, to reconstitute all colours correctly, it is at least possible to obtain in this way very pleasing effects which can even give an almost perfect illusion of reality where there is no means of comparison, as is the case when pictures are projected in a darkened room. It is of course necessary that the missing colours should not be essential to the picture (L. I)ucos du Hauron, 1895).
This incomplete method of colour reproduc tion has been, until quite recently, the one most generally used in colour cinematography, either with additive (alternate view ing of the elementary images ; Kinemacolor process of G. A. Smith, 1906), or with the sub tractive method (elementary monochromes, printed in register on both sides of the film ; J. E. Thornton, 1912 ; Technicolor, and other processes).' The colours chosen are usually a vermilion red (colour of the normal red filter used in three colour selection) and a greenish-blue (colour of the greenish-blue pigment of three-colour subtractive synthesis). The negative exposed under the red filter is used for producing a greenish-blue monochrome, and the negative exposed under the greenish-blue filter is used for a red monochrome. As a matter of fact, these two colours do yield a satisfactory render ing of flesh tints. It is obvious that in this way it is not possible to obtain either yellow or purplish-blue, so that this pair of complemen taries is but ill-suited for landscape photography (unless the sky does not appear in the picture). 2 There have been several suggestions or patents for reducing to two the number of the selection negatives for three-colour synthesis, the third negative being obtained later by combinations between one of the original negatives and a transparency printed from the other negative. Experience has shown that none of the proposed combinations produces the expected result, and simple reasoning will show that such hopes are chimerical.