910 Dr. Marine Siegbahn on the Use of the 



above-mentioned paper, the apparatus bare been so ranch 

 improved that essentially better registrations can be obtained, 

 and the measurements can also be made more accurately. 

 Concerning another method, which may possibly give the 

 same result, see Gruyau, C. R. 1913, p. 777. 



The instrument employed can be seen in fig. 1 (PI. XVI.). 

 It consists of two coaxial microscopes, the first o£ which 

 gives a greatly diminished (tJ-q) image of an illuminated slit 

 (1 mm. broad) . To the telephone-membrane is attached a sharp 

 edge, which is placed in the image so as to screen off a great 

 portion o£ the light (9/10). The rest of it will pass through 

 the other microscope and form a magnified image on a 

 photographic plate. While the membrane swings, the edge 

 will screen off more or less of the first image, which will 

 cause a corresponding variation in the light-intensity of the 

 image on the plate. On moving the plate perpendicular to 

 the ray of light, there will be sections with different bright- 

 ness, which will give dark or bright lines on developing the 

 plate. 



The darkness of the plate is a measure of the corresponding 

 deviation of the membrane. In order to give a complete 

 representation of the membrane-vibrations, the darkness of 

 the plate must be continually registered. In the above- 

 mentioned paper I have used two methods for that purpose. 

 In the following investigation a variation of these two 

 methods has been employed. The image of a Xernst-lamp- 

 pin was thrown on the plate by means of a cylindrical lens ; 

 part of the light will then pass through the plates and fall 

 on a thermopile. The produced thermoelectric force was 

 measured by a self-registering galvanometer (from Hartmann 

 <fc Braun). On moving the photographic plate slowly in its 

 own direction the galvanometer registers a curve, where the 

 deviations are approximately proportional to the degree of 

 the darkness of the different sections, that is ; proportional to 

 the deviations of the telephone-membrane. 



There will be faults in the registered curve arising out of 

 the following three factors : — 



1. The breadth of the light-line thrown on the plate. 



2. The slowness of the thermopile. 



3. The free vibrations of the galvanometer-coil. 



We can understand the effect of these causes, by thinking 

 of a plate with a sudden transition to dark. In letting this 

 section pass over the light-line we shall obtain a slow transition, 

 instead of a sudden one ; otherwise, we may receive a 



