ASTRONOMY 183 



effect of distortions in the gelatine during drying was considered 

 in the second paper, but there are other causes which may 

 affect the relative positions of adjacent images. The study of 

 these is of utmost importance in the modern applications of 

 photography to the determination of the relative positions of 

 double stars and satellites, and to the measurements of adjacent 

 spectral lines. The results obtained from previous investi- 

 gations have been somewhat contradictory, at some times ap- 

 parent attractions and at other times apparent repulsions having 

 been indicated. The explanation of these discrepancies is to 

 be found in the complex nature of the phenomena. There are 

 three distinct phenomena concerned, (i) Owing to the turbidity 

 of the emulsion there is a mutual light action in the case of 

 two adjacent images. For two equal circular sources the out- 

 lines of neighbouring images, as the exposure is gradually 

 increased, are a series of lemniscates which for sufficiently 

 small exposures consist of two distinct ovals. The centres of 

 these ovals are displaced towards each other relatively to 

 the centres of the geometrical images, and the displacement 

 increases with the exposure. Therefore, on this account, 

 double stars and close spectral lines are subject to an apparent 

 attraction whose amount depends upon the separation of the 

 adjacent borders of the two images. This effect is negligible 

 except for very small images, which are very close together, 

 (ii) Actual displacements of the gelatine : it has been seen 

 that these produce an apparent attraction between images, 

 which can be brought into evidence by measuring the plate 

 when wet and when dry. Where it is possible that such effects 

 may enter and become of importance, they can be almost 

 completely avoided by suitable choice of developer, (iii) The 

 third cause of apparent displacement is a developer effect. 

 During the process of development, reaction products are 

 formed which have a restraining action on the development. 

 Now consider a pair of adjacent images : as the development 

 proceeds these reaction products are formed around both 

 images and gradually diffuse to adjacent regions. In the space 

 between the images these products enter from both sides, and 

 their concentration is therefore greater than on the far sides of 

 the images. The development at the adjacent edges is therefore 

 retarded, with the result that there is an apparent repulsion 

 of the centres of the images. This effect is present in varying 

 degree with all developers, but is greatest with pyro and least 

 with metol-hydroquinone. It increases with the timeof exposure. 

 The net result of the three effects is that with over-exposed 

 images there is repulsion, due to the large developer-effect, 

 whilst with normal exposures a slight attraction is probable 

 but the errors should not exceed two or three microns* 



