364 SCIENCE PROGRESS 



these plates must be regarded as a triumph of manufacturing 

 skill. Some very excellent results have been obtained with 

 them, but the colours are readily affected by error in their treat- 

 ment. The "grain" of these plates is not so fine as one might 

 suppose from the fact that starch granules are used, for it is 

 apparently impossible to thoroughly mix the differently coloured 

 granules. They occur on the plates in groups up to a dozen or 

 more of one colour, and the groups are visible by a magnification 

 much too low to show the individual granules. 



A one-plate process such as that just described obviously 

 must have the same dyes for taking the photograph, that is for 

 dividing the colour of the original into the three fundamental 

 colours, as for viewing the result, and this is a disadvantage. 

 It is due to the wide permissible choice of colours and the 

 non-critical comparison of the resulting colours with those of 

 the original that the results are acceptable. These plates re- 

 produce the colours of the spectrum very poorly. Presumably 

 in the selection of dyes the best compromise has been made for 

 such photography as the plates wall generally be used for. It is 

 practically impossible to multiply such colour photographs by 

 exposing a new plate to light under a finished plate, because, as 

 we have seen in the first plate, two-thirds of the light is lost, 

 and the effect of making such a cop}'' would be to lose two-thirds 

 of the light passing through the first. Such a copy would, 

 therefore, have about eight-ninths of its surface dark, reducing 

 the light that would otherwise be available to about one-ninth, 

 and it would be so dark as to be almost useless. Mr. Powrie 

 has shown how it is possible to overcome the difficulty when 

 using his plates, by fixing a mirror on each side of the printing- 

 frame so that the light impinges on the plate from three direc- 

 tions, and then by suitably separating the two plates by a sheet 

 of glass or celluloid every line in the upper plate — the picture — 

 can be made to give a line of three times its width on the lower 

 plate. So each of the three colours in the upper plate covers 

 the whole surface of the lower plate, and no light is lost. By 

 using light that corresponds to one colour only, it is possible by 

 this means to get a record of that colour on an ordinar}' plate 

 that is continuous — that is, that covers its whole surface instead 

 of being as the original is, in lines that cover onl}' one-third 

 of its surface. Three separate continuous colour records, one 

 for each colour, can thus be obtained from the original single 



