296 INFRA-RED EMISSION SPECTRA. 



tremors, even with the most sensitive arrangement used in the vacuum- 

 tube radiation. Those v^ho have used sensitive galvanometers with 

 their bolometers have been obliged to provide elaborate suspensions to 

 protect the instrument against earth tremors and confine their observa- 

 tions to the quiet hours of night. The chief disadvantages in a radio- 

 meter of great sensitiveness, having linear vanes, is its long period, due 

 to the viscosity of the residual gas. In fact, viscosity is the chief prob- 

 lem to contend with in constructing a very sensitive radiometer having 

 a short period. In a bolometer the spectrum is projected directly upon 

 the bolometer strip, which can be made very narrow, while the period 

 of the accompanying galvanometer is not to such a great extent deter- 

 mined by viscosity of the air. In the radiometer, since the vanes re- 

 volve about an axis, the spectrum must be projected upon a stationary 

 slit, back of which one of the vanes is exposed. Now, the width of the 

 vane that must be used depends entirely upon its distance from the slit. 

 The slit is in the vertical focus of the spectral lines, which wiil diverge 

 again after passing through it into the radiometer. In the original 

 Nichols radiometer' the slit was at least 3 cm. from the vanes, which, 

 as a consequence, had to be about 2.5 mm. wide and 1.5 cm. long. For 

 his work on the radiation from stars^ no slit was required and the size 

 of the circular vane was made equal that of the star image, viz, 2 mm. 

 diameter. This eliminated viscosity to a great extent, and he succeeded 

 in obtaining a period of 11 seconds. For line spectra the omission of 

 the slit would hardly answer the purpose where great accuracy in the 

 position of the wave-length is concerned. In the latter case the prob- 

 lem is to place the vane as near the window as is possible. This can be 

 done by using but one window,' as shown in figure 133. 



The use of but one window introduces a new complication in that 

 any slight change in its temperature will immediately affect the radio- 

 meter vane, i. e., will cause a shift in the zero reading. This was par- 

 ticularly noticeable in the arc-spectra work, where the radiation is very 

 intense. Consequently a suspension was used which gave a maximum 

 deflection in 8 seconds. This was less sensitive than for the vacuum- 

 tube work, but the arc lines are very intense and it answered the pur- 

 pose, since exposing the vane for a longer time would warm the win- 

 dow and cause a large shift of the zero reading. On the other hand, 

 for the vacuum-tube radiation, where the radiation is very weak, the 

 radiometer must be of extraordinary sensitiveness, to obtain which a 



^Nichols: Phys. Rev., 11, p. 891, 1897. 

 ^Nichols: Astrophys. Jour., 13, p. loi, 1901. 



