248 SCIENCE PROGRESS 



beyond the resolving power of almost all grating, and certainly- 

 all prism instruments. Michelson, and later Fabry and Buisson, 

 applied the interferometer to measurements of this nature. 

 The difference in wave-length corresponding to successive fringes 

 in any form of interferometer diminishes as the order of inter- 

 ference is increased. Now we have seen that a spectrum line 

 is not strictly monochromatic, so the light from different parts 

 of the line will overlap as the order of interference is increased, 

 and finally uniform illumination will result. The number of 

 fringes visible before this happens bears a known relation to the 

 breadth of the spectrum line, and by counting them Michelson 

 actually deduced the half-widths of a number of lines. This 

 application of the interferometer has produced very valuable 

 results, but it suffers from its dependence on personal estimates 

 of the point at which the fringes become invisible, and the 

 practical difficulties are great when dealing with faint and 

 closely packed spectrum lines. For these reasons, half-width 

 measurements in the secondary hydrogen spectrum are un- 

 satisfactory if made by this method, and the half-widths of 

 a few lines have now been redetermined by Prof. Merton and 

 the writer, using an entirely new method, with results pointing 

 conclusively to the molecule of hydrogen as the source of the 

 spectrum. 



The principle underlying the new method is the existence of 

 a " critical photographic intensity." The density of the 

 photographic image of a source of light, for a fixed time of 

 exposure, is not proportional to the intensity of the light, but 

 decreases much more rapidly than does the intensity, owing to 

 the peculiar " law of blackening " discovered by Huerter and 

 Driffield. If an extended source of light is taken, in which the 

 intensity falls off gradually to zero on either side of a maximum, 

 the photographic image will not fall off in intensity in quite the 

 same way. It will have a fairly sharp boundary on either side, 

 where the incident light has diminished to a critical intensity 

 below which it is unable to afi"ect the plate, and above which it 

 produces a comparatively dense image (the absolute value of 

 this critical intensity depends of course on the time of exposure). 

 The sharpness of the boundary can be increased by using special 

 plates made to give strong " contrast," by developing in strongly 

 alkaline hydroquinone, and by chemical treatment subsequent 

 to development. A spectrum line under a sufficiently high 

 dispersion forms just such an image, owing to the Doppler 

 broadening, and the width of this image depends on the distance, 

 on either side of the central maximum, at which the intensity 

 has dropped to the critical value. Suppose two successive 

 photographs of a line are taken on adjacent parts of the same 

 plate, and are given the same time of exposure, and the same 



