DESCRIPTION OF APPARATUS AND METHODS. 23 



one direction as the temperature of the room increased, due to the pres- 

 ence of intense radiation from the Nernst " heater." 



The drift lasted only a short time, in beginning a series of observa- 

 tions, and after that the shifting for a single reading was rarely greater 

 than 0.1 mm., while generally for a period of 10 to 12 minutes no shift- 

 ing at all could be detected. 



As a consequence the deflections were always trustworthy, even at 

 14 fi, where they were only from 2 to 3 mm,, while in passing through 

 absorption bands in this region they were often only from o.i to 0.2 mm. 

 Since the rate of increase of a deflection with the time of exposure of 

 the vane to heat follows a logarithmic curve, which increases very rap- 

 idly at first, then more and more slowly, and finally being asymtotic, one 

 can make a reading, in the region where the deflections are small, in a 

 shorter time than the actual period of the instrumient, thus avoiding 

 a possible drift. 



The radiometer slit was mounted upon the spectrometer arm, so that 

 if any shifting occurred in the relative positions of the radiometer and 

 spectrometer^ it would not affect the location of any point of the spec- 

 trum with respect to the slit. This is of importance in measuring the 

 distribution of the energy in the spectrum of the radiator, but in this 

 work we are dealing with the ratio of the intensity of the radiation 

 which has passed through an absorbing medium to the intensity of the 

 direct radiation, in any region of the spectrum, so that we are not con- 

 cerned with the change in the sensitiveness of the radiometer, which 

 varies from day to day, and it was sufficient to know that the sensitive- 

 ness did not change while determining this ratio. This is more likely 

 to be afifected by variation in the intensity of the source of radiation, 

 which will be noticed in discussing that subject. 



THE ABSORPTION CEI.LS, 



One of the chief difficulties to contend against in this work is to obtain 

 pure chemicals, and it is of the greatest importance to prevent contami- 

 nation while investigating them. To this end a suitable absorption cell 

 had to be devised for containing the liquids. The cell walls were made 

 by splitting the rock-salt crystal parallel to a cleavage plane. This gave 

 thin plates that were quite plane, smooth, and of a finer polish than 

 could be obtained by hand polishing. Furthermore, the surfaces are 

 not so easily attacked by moisture. One form of cell, used in examining 

 the low boiling-point liquids, consisted of a fine wire, from o.i to 0.3 mm. 

 in diameter, which was covered with Le Page's glue, pressed between 



^Stewart, loc. cit., experienced this difficulty. 



