192 
From the mean value of 2 from table II, it would appear that the 
value does not vary from unity by more than one-tenth of cne per cent. 
An unfavorable combination of errors could affect the single values by 
three-tenths of one per cent. 
Tube A, in Steam. Tube B, in Ice Bath. 
Exp. 7. meas. A cor. ‘Te abs: A meas. / cor. T. abs. =~ 
1 38.287 38.325 370.80 32.862 . 32.867 | 272.5 | 1.000758 
2 38.262 mf 38.299 " 370.76 32.838 “| 32.842 i | 1.000485 
- 3 38.303 - 38.339 371.12 32.860 af 32.864 cy, a 1.000707 
4 | 38.234 38.271 370.49 32.846 | 32.851 | | 1.001770 
eae 38.252 ae 38.288 i 370.67 32.843 | 32.848 | 1.001180 
6 | 3g275 38312 | 370. 87 ~~ 39.807 | 32.811 | 0.998216 
7 38231 | 38.268 370.84 | | 32.863, 32.863 \ 32.869 > Sa 1.003972 
8 a “38.291 “| 38.328 ¥ 370.91 sion | | 3 ae s 32.869 5 ee 1.001020 
oy ¢) eee aes eee sa mean 1.001013 
= 0.000376 
Table 2. 
Procedure, Series II. For the second series of measurements the pro- 
cedure was substantially the same as that for the first. Carefully dried 
and purified nitrogen was introduced into the tubes. The upper tube sur- 
rounded by cotton and enclosed in the double walled jacket, was allowed to 
assume the temperature of the room, its thermometer being read through 
the mica windows. ‘The lower tube, 2.2 cm. in diameter, was immersed in 
the liquid air bath, the top of the tube being 2 or 3 cm. below the surface. 
Temperatures of the liquid air were read by means of a constantan-iron 
thermo-junction and a sensitive millivoltmeter, which was provided with 
a calibration curve from the Reichsanstalt. These temperatures were 
checked by evaporating samples of the liquid air, mixing with hydrogen 
and exploding by means of an electric spark in a eudiometer tube. From 
percentages of oxygen thus found temperatures were interpolated from 
Baly’s curves." 
4 Phil. Mag. 49, June 1899. 
