THE COMBINING VOLUMES OF HYDROGEN AND OXYGEN. 
415 
apparatus from the warmth of the observer’s body, a sheet of asbestos was fastened 
in front of the manometer and tubing below the dead-space, with the result that the 
cooling effect of the ice-baths above came more into play, and the temperature 
registered by the thermometer was lower the longer the system had been in ice. 
Consequently the hydrogen fillings, which were made first in the day, contained 
relatively too little gas. The result of applying the correction to series 4 was to 
lower the mean by 1 part in 50,000. 
The following table gives the results of series 4 : two additional columns, 6 and 7, 
give the values for the ratio and the atomic weight when the small correction is 
applied. 
Table IY. 
1. 
Hydrogen 
excess. 
2. 
Hydrogen 
residue. 
3. 
Difference. 
4. 
Ratio of 
combining 
volumes. 
5. 
Atomic 
weight of 
hydrogen. 
6. 
Ratio of 
combining 
volumes. 
7. 
Atomic 
weight of 
hydrogen. 
c.c. 
c.c. 
c.c. 
1 
2-443 
1 • 527 
0-916 
2-00298 
1-00778 
2-00300 
1-00779 
2 
3-717 
2-827 
0-890 
2-00290 
1-00773 
2-00284 
1-00771 
3 
2-846 
1-903 
0-943 
2-00307 
1-00782 
2-00299 
1-00778 
4 
3-589 
2-614 
0-975 
2-00318 
1-00787 
2-00305 
1-00781 
5 
3-453 
2-507 
0-946 
2-00308 
1-00783 
2-00309 
1-00783 
6 
3-311 
2-410 
0-901 
2-00293 
1-00775 
2-00288 
1-00773 
7 
3 • 895 
2-945 
0-950 
2-00309 
1-00783 
2-00306 
1-00781 
8 
2 • 945 
2-015 
0-930 
2-00303 
1-00780 
2-00301 
1-00779 
9 
2-015 
1-052 
0-963 
2-00314 
1-00785 
2-00307 
1-00782 
10 
3-448 
2-521 
0-927 
2-00302 
1-00780 
2-00310 
1-00784 
Mean. . . . 
2-00304 
1-00781 
2-00301 
1-00779 
±0-00002 
±0-00002 
Mean deviation, 1 in 
7000. 
The mean value of the ratio in column 6 differs by only 1 part in 100,000 from the 
final mean of the first three series when this is corrected for the imprisoned bubble. 
It may be noted in the above table that without experiments 2 and 6 the 
maximum deviation in column 4 is 1 part in 10,000, and in the corrected values in 
column 6, only 1 part in 18,000. 
The explanation of the two low results is to be found in another temperature error 
which affects all four series. The final setting of a determination had sometimes been 
repeated on the following day, with the object of testing the accuracy of the measure¬ 
ment. On such occasions it was noticed that the volume of gas in the pipette had 
apparently increased, unless the bath had been iced for some time before the reading 
was made. It was therefore necessary to make a careful investigation of the time 
