THE RADIOMETER. 



299 



Numerous shields were provided to prevent stray radiation from enter- 

 ing the radiometer. The data below the line in the center of the table 

 show the constancy of the inverse square law which would indicate that 

 within experimental errors the deflections are due to radiation from the 

 candle onh'. 



Table II. — Sensitiveness of Radiometer. 



[Deflections are in centimeters per square millimeter of exposed surface (viz, i sq. mm.) 

 for a candle and a scale each at a distance of one meter.] 



In figure 134 is given the energy curve of acetylene for the visible 

 spectrum, and that of a Nernst heater for the region of 14 ^u, for the 

 very sensitive radiometer used under similar conditions in the vacuum- 

 tube work, curve a, and, for the old radiometer used in the absorption- 

 spectra work, curve Z?— ordinates are deflections in centimeters, radio- 

 meter slit 0.7 mm., spectrometer slit i mm. The sudden drop from 

 30 cm. deflection at 13 /^ to 2 cm. at i6/x is due no doubt to the opacity 

 of the rock-salt prism, which is very opaque in this region. 



Radiometer construction is still in its infancy. Other improvements 

 suggest themselves. For example, there is no special reason for having 

 two vanes, for symmetry, when using a torsion-head. B}' using a metal 

 counterpoise instead of one of the vanes, the viscosity effects would be 

 reduced by almost one-half, hence the period shortened. 



Another form, which is a combination of a radiomicrometer, which 



