3IO INFRA-RED EMISSION SPECTRA. 



is due to gases absorbed by the electrodes. At any rate it did not inter- 

 fere with the work, since the tube was used at such a pressure that it 

 did not become heated. All evidences indicated that the impurity band 

 at 4.75 IX., which was found in all the gases in the preliminary work, was 

 due to something that entered with the gas ; and it was finally shown to 

 be due to contamination with CO2 while making the gas. The problem 

 then was to make a gas which was free from even slight traces of CO2. 

 Previous exeperience showed that to bubble a gas through several puri- 

 fying solutions in series with the pump was not sufficient. Hence, in 

 preparing such a gas as CO, an ordinary mercury gas pipette was 

 used. The gas was washed back and forth from the mercury pipette, 

 through a tube containing P2O5 into a pipette of KOH for about half 

 an hour, to free it from CO, and to dry it. After such a treatment 

 nitrogen and oxygen did not show the impurity band at 4.75 jn. 



The residual gas in the pump and glass connecting tubes was swept 

 out by passing a discharge through it and heating it from the outside. 



A large Charpintier (said to have 125 miles of wire for a secondary) 

 and also a Max Kohl No, 4 induction coil were used to excite the 

 vacuum-tube. For convenience, for most of this work, the primary was 

 connected directly through a rheostat to a 104- volt, 60 or 120 cycle 

 alternating current. 



An ammeter was placed in the primary circuit, while an a. c. milli- 

 voltmeter having a resistance of 21,760 ohms was used to measure the 

 current in the secondary. 



The primary current varied from 3 to 6 amperes, and the secondary 

 from 0.012 to 0.028 ampere. 



Since the spectrometer arm was movable, the vacuum-tube had to be 

 adjusted before the slit, by hand. The points on the curves are usually 

 the mean of several readings. 



Observations were made for constant current in secondary and vari- 

 able pressure, and also for constant pressure and variable current. 



The method of observation consisted in obtaining the radiometer 

 deflection for the gas when the discharge was passing and then, after 

 stopping the discharge, allowing the deflection to return toward its 

 original zero reading. The hot cell would prevent the deflection from 

 returning to the original zero in 50 seconds (radiometer period) and 

 the difference in the two zero readings gave approximately the deflec- 

 tion due to the hot cell. Another method for finding the deflection due 

 to the hot cell consisted in passing the discharge for 50 seconds, then 

 on stopping the discharge and raising the shutter, reading the deflection. 

 Of course the cell cools somewhat while obtaining the deflection, but 



