Aprils, 1898.] 



scmncR 



All 



pear-shaped glass bulb B, about 112 mm. 

 in diameter. The stem of the thermometer 

 hangs freely in the long neck of the large 

 bulb. I shall hereafter call the glass bulb 

 B the ' large radiation bulb,' or simply the 

 ' large bulb,' to distinguish it from a smaller 

 one used later. The bulb B is surrounded 

 by a copper tank C, lagged with woolen 

 cloth, and filled with crushed ice and dis- 

 tilled water. A wire netting C serves to 

 keep some of the ice always below the low- 

 est point of B. The tank C is movable on 

 vertical guides, whereby it may quickly be 

 raised to, or lowered from, the position 

 shown, thus exposing the bulb B alternately 

 to the ice bath and the atmosphere of the 

 laboratory. The bulb B communicates 

 freely with the large barometer tube D, 

 which is used for measuring all but very 

 small pressures. E is a standard boiled 

 barometer, dipping into the mercury cistern 

 F, common to both barometers. G is a 

 McLeod gauge giving very accurate meas- 

 urements of small pressures, and H is a 

 drying bulb containing phosphorous pen- 

 toxide. The glass stopcock I serves to ad- 

 mit other gases than air. The mercury 

 valve K prevents any leakage backward 

 from the pump when the latter is stopped, 

 during observations. Exhaustion is effected 

 by an automatic Sprengel pump having five 

 fall tubes. L is a fine ^athetometer placed 

 in front of the whole apparatus, and by ro- 

 tation on its vertical axis is adapted to read 

 the McLeod gauge, both barometers, and the 

 thermometer. It has a vertically divided 

 scale with vernier and microscope, for read- 

 ing the barometers, and a micrometer for 

 reading the gauge. A watch N is mounted 

 close beside the thermometer on a sliding 

 frame, so as to be easily kept in the field of 

 view of the cathetometer telescope when 

 the latter is used to observe the falling 

 temperature. 



Before using this apparatus, I always 

 exhausted to a good vacuum and heated 



the bulb B by means of a water-bath, and 

 all other vacuous parts by means of an air 

 bath, to 100 degrees for several hours. This 

 was found necessary in the first instance 

 with air, in order to divest the inner glass 

 surfaces of that portion of their coating of 

 adherent gas most easily given off in a 

 vacuum. This gas was pumped out, and, 

 not being principally air, was not largely 

 reabsorbed when air was admitted. With- 

 out this precaution I was unable to obtain 

 constant results at very low pressures. 

 When other gases were tried successively, 

 the preliminary heating prevented gas from 

 one operation attaching itself to the glass 

 and remaining to contaminate the succeed- 

 ing 'gas at very low pressures. 



I next introduced the proper gas up to 

 atmospheric pressure and made a prelimi- 

 nary cooling of the thermometer by raising 

 the ice tank C. This preliminary cooling 

 was found to have a slight effect on the 

 readings next following, and was done to 

 make the first set of readings on any day 

 entirely comparable with the others. I 

 then lowered the ice tank, and, when the 

 temperature had raised to 18 degrees, 

 stirred the ice and water thoroughly, raised 

 the tank again, and observed the thermom- 

 eter through the telescope — noting by the 

 watch ISr the instant when the falling mer- 

 cury passed each degree of the scale. Then, 

 with the ice tank still up, I noted the pres- 

 sure by measuring with the cathetometer 

 the difiMrence in height of the barometer 

 columns in D and E. The barometer D 

 showed that the gas in the radiation bulb 

 cooled nearly to zero with very great ra- 

 pidity when the ice tank was raised. I 

 always measured pressures with the radia- 

 tion bulb cold. It was usual to repeat the 

 whole operation to confirm results before 

 reducing the pressure by the pump. 



Observations were thus made at pressures 

 varying from atmospheric down to the best 

 vacuum obtainable. In some instances 



