APPENDIX I. 121 



less than 3 per cent of SiOa. The examination of the pure feldspar 

 radiator was not successful, the platinum breaking soon after beginning 

 the observations, and the examination was never repeated. Whether 

 the platinum was acted upon by the feldspar is not known. The heater, 

 when covered with ordinary blue clay, gave a curve quite similar to 

 curve a, the only difference being slight depressions at 6 /i. Since some 

 of these curves were obtained at different periods, several months inter- 

 vening, the relative intensity of the radiation from the different heaters 

 can not be compared, except curves a and b in the two figures. They 

 are given for the purpose of showing the relative distribution of the 

 energy in the different spectra. 



Curve d in fig. 126 shows the radiation from a new heater, using the 

 large spectrometer, which had mirrors of i meter focal strength ; con- 

 sequently the radiation from the heater had to traverse almost 3 meters 

 of air before entering the radiometer. The result is a marked trans- 

 formation of the whole curve. The water- vapor bands at 2.9, 4.8, 5.1, 

 5.8 to 6.0, and at 6.6 /a and the CO2 bands at 2.7 /x and 4.28 /* have been 

 determined by Paschen (loc. cit.). The source of the 3.45 /u, band is 

 not known, unless it be due to the presence of hydrocarbon vapors, 

 which manifest themselves, just as Pauer (loc. cit.) found for the ultra- 

 violet. As far as known to the writer this is the next to the largest spec- 

 trometer ever constructed for such work (Langley's being the largest) , 

 and it illustrates well the difficulties to be encountered in increasing the 

 dispersion by lengthening the spectrometer arms. For example, the 

 intensity of the radiation drops from 21 to 8 in the CO2 band at 4.25 fi, 

 which shows that a very much more intense source of radiation will be 

 necessary if a larger spectrometer is to be used. What is needed, then, 

 is a device giving larger dispersion, with short spectrometer arms. For 

 example, a fluorite prism on the small spectrometer would have given 

 about the same dispersion as this large spectrometer and rock-salt prism. 



