248 



Dr. IMuller on the Thermal Effects 



spectra are desired. On bringing the lineal thermo-battery 

 (rather more than 2 millims. in width) into the place of the grid, 

 the total action of the pencil of rays falling directly upon the 

 thermo-battery was a deflection of the multiplier of 30°. If, now, 

 the same quantity of rays fall upon a portion of the grid 

 2 millims. in thickness, about half of the rays are intercepted by 

 the opake portions of the grid ; consequently a total thermal 

 action of 15° remains. This heat, however, is distributed among 

 a whole series of spectra. If, neglecting the middle image, we 

 only suppose there to be seven such spectra, scarcely 2° remain 

 for the total thermal action of a single refraction-spectrum, even 

 neglecting all other sources of loss. But if the total eflfect of 

 such a spectrum is only 2°, it is impossible to expect noticeable 

 thermal efi"ects in its separate portions. 



Although experimental means for accomplishing this purpose 

 are not yet so complete as to admit the derivation of the course 

 of the thermal cui-ve in the diffraction-spectrum by means of 

 direct experiments, still it is possible to get at the same result 

 in an indirect manner. The course of the thermal curve in the 

 diffraction-spectrum may be derived from that of the thermal 

 curve obtained (in fig. 1, p. 243) for the refraction-spectrum, 

 most simply in the following manner : — 



In fig. 2, p. 245, let us imagine perpendiculars drawn to 

 the line K. S from the points in which the horizontal lines cut 

 the curve ab c. These perpendiculars will divide the whole 

 length of the spectrum into fifteen divisions, each of which corre- 

 sponds to a difference of 0-0001 millim. in the undular length ; 

 that is to say, passing from left to right, every successive ver- 

 tical corresponds to an undular length greater by O'OOOl millim. 

 than the preceding one. Fig. 3 represents the curve of thermal 

 Fig. 3. 



intensity for the refraction- spectrum, together with these verticals. 



If, now, a diffraction- spectrum, of a width H S, equal to that 



of the refraction-spectrum (fig. 3), has to be divided into fifteen 



parts, each of which corresponds to an increase of 0-0001 millim. 



