APPENDIX III. 125 



window with great tenacity. After standing several days the vanes 

 no longer become electrified so easily. 



Since the completion of this research the radiometer has been remod- 

 eled, in order to gain a greater sensitiveness, and at the same time to 

 shorten the period. It became evident that on account of the viscosity 

 of the residual gas in the radiometer the period is more affected by the 

 size of the vanes than by the moment of inertia of the system or the 

 diameter of the fiber suspension. In order to decrease the area of the 

 vanes they must be situated near the slit of the spectrometer. To this 

 end the outer window, R^, was discarded, and only the inner one, R^, 

 was used (fig. 3). It rested on a wide flange, and was made airtight 

 by means of beeswax covered with shellac varnish. The slit F,, of 

 the spectrometer (fig. i) was mounted directly upon the outside of this 

 window, while the vanes were about 2 mm. from the inside surface. 

 To prevent the spectrometer from slipping, with respect to the radio- 

 meter (and the slit F), it was held in place by means of heavy weights. 



The vanes v/ere of very thin strips of mica, 12 mm. long and i mm. 

 wide, secured to fine glass rods. The mirror was about the size of the 

 preceding. Unfortunately, the weight the suspension was not deter- 

 mined, but since the area of the vanes, instead of their mass, concerns 

 us most, that is of minor importance. 



The half period, i. e., time for maximum of deflection, of this system 

 was only 6 seconds, and at 0.05 mm. pressure its damping was so small 

 that it was used ballistically. The sensitiveness was about 4, i. e., a 

 paraffin candle at a distance of i meter gave a deflection of 4 cm. per 

 square millimeter of exposed surface (slit o.i mm.) upon a scale situ- 

 ated at a distance of i meter. This was a somewhat greater sensitive- 

 ness than the preceding, and was equal to that used by Stewart (loc. 

 cit.)j which had a period of 40 seconds. A still greater sensitiveness 

 was attained by selecting a much finer fiber which was aperiodic in air. 

 In comparison with the bolometer this does not seem a fair test of sensi- 

 tiveness, for the large vanes of the Stewart radiometer, with a new 

 fiber (period 20 seconds), had a sensitiveness of only i instead of 4. 

 Nevertheless, the dispersed radiation from the hand was sufficient to 

 give a deflection of from i to 2 mm. at 9 11. 



For work where a very narrow linear absorbing surface is required 

 it seems quite probable that the sensitiveness of the radiometer can not 

 attain that of the bolometer. Nevertheless for ordinary laboratory 

 work it seems superior to it on account of its simplicity and its freedom 

 from magnetic and thermal disturbances. The radiometer, as ordi- 

 narily used, is from 4 to 8 times as sensitive as the Boys radiomicrom- 

 eter, which was thought capable of detecting a rise in temperature of 

 0.000001° C. 



