THIS process was first suggested by Chas. Cros in 1869. He proposed to split up the light by means of a prism, which was to be turned so as to direct the three colored rays to slightly different positions on the sensitive surface. Cros' idea was purely theoretical and he made no attempt to carry the process into practice. The idea was subsequently taken up by J. Drac and L. Moelants, but has never passed the experimental stage. Subsequently F. M. Lanchester, in 1894, patented a modified process in which, instead of using a single slit to obtain a single spectrum, which obviously considerably limited the available light, a series of slits, that is, a black and white line screen, was used, so that a series of minute spectra would be formed on the sensitive surface. Assuming that the light incident on these slits was reflected from a colored object, it is clear that only the colors of the object would appear in the individual spectra. Various modifications of this process were proposed by Raymond, Cheron, Lippmann and others. The disadvantages of all these methods was that the image of the object and that of the slits must be simultaneously projected on the sensitive surface; consequently the apparatus was extremely bulky, and the available light very little. They also suffered from the obvious disadvantage, incident to all prismatic spectra, that the distribution of the colors is very unequal, the reds being cramped together and the blues unduly spread out. Thus one may roughly state that the blue rays would occupy about three times the space of the red, instead of being proportionately distributed, as in normal or diffraction grating spectra.

In 1904, J. Rheinberg devised an apparatus in which this unequal distribution of the colors was overcome by the use of a specially computed direct-vision prism, with which the color spaces were practically the same as in a normal spectrum. This, like all these processes, is better fitted for the laboratory than for practical work, as the original taking apparatus has to be used for viewing the results. There is no commercially available instrument and it would be somewhat costly to make. Full details will be found in The Photographic Journal, 1912: 162, and The British Journal of Photography, 1912, Color Photography Supplement, 19, 28, 33, 38.

Another process, which might be called the refraction process, was first suggested by Liesegang in 1896, and a year later independently by J. A. C. Branfill. In this process the action of the cross-lined screen is taken advantage of, namely that each aperture in a cross-lined screen acts as a pinhole camera and produces an image of the diaphragm. Thus, when using a square diaphragm, the dots on the sensitive surface are square, and with triangular stops three cornered figures are formed. If the diaphragm, instead of being a single aperture, is divided into a series of apertures, there are as many dots on the sensitive plate as there are apertures. This being established, it is clear that if the stop aperture be split up into three areas, each covered by a color filter, one would have three images formed in the three colors, consequently the image of a colored object transmitted by a lens would be represented on the sensitive surface by minute areas corresponding to the colors of the subject. One may look upon this method as a screen-plate process in which the screen-elements are optically formed during exposure and not on a separate, preformed plate.

There is considerable loss of light in this process, due to the fact that the apertures in the cross-lined screen transmit but a fraction of the incident light. The apparatus necessary is only the cross-lined screen, which presents no particular difficulty, as it can be obtained commercially. The only requirements to be satisfied are that the opaque lines shall preferably be the same width as the transparent interspaces, and that the lines cross at right angles. The fineness of the ruling is elastic, as one may obviously decide to work with a fairly coarse or a fine ruling, according to the purpose of the final picture. Probably for general work from 70 to 100 lines per inch would suffice, as this would give color elements one-third the size, that is, from 210 to 300 per inch. Theoretically, there is no limit to the fineness, but the finer the ruling the more difficult it becomes to register the picture.

The construction of the diaphragm is not an easy matter, for one is naturally limited as regards the size of the filters by the working aperture. Consequently one must decide always to use a given aperture or else to prepare as many filter diaphragms as may be necessary to satisfy the requirements of sharpness of the image, taking into consideration here the marginal definition on the plate. Obviously the most practical plan would be to use as large a diaphragm as possible, and, drawing a circle on paper of the exact size, to describe a square outside the circle and divide this into equal areas, which shall be those of the filters, trusting to a supplementary filter to obtain equal exposures. Or one might vary the areas of the filters corresponding to the filter factors, but then one must have unequal color areas in the resultant picture; in all probability this would be the most satisfactory plan, as this will theoretically be necessary to form a neutral grey. These are points which would have to be worked out, as no one has done more, so far as the author is aware, than to work this process on paper.

The only possible plan for making the filter diaphragms would seem to be to use a microscopic cover glass, of the required size to fit the lens tube, and cement the filters down in strip form, which is not such an easy matter, as the edges must fit without white interspaces or overlapping. Or one might expose on bichromated gelatine and stain up with dyes that take differentially on hard and unhardened gelatine; when one would have the difficulty of adjusting the color depth to the rigid requirements of the theoretical filters. One also meets here with the difficulty of inserting the diaphragms in the lens tube, as it is quite possible that a little trouble might ensue in the case of anastigmats of large aperture because of the thickness of the glass, small though this is.