420 
DR. F. P. BURT AND DR. E. C. EDGAR ON 
contraction curve was found to be the same whether this portion of the manometer 
was at room temperature or at 0° C. As the bath was used in every gas measurement, 
the effect of the resulting small increase in the quantity of gas required to fill the 
pipette cancels out. 
The possibility of a small error, due to the difference of procedure in the two final 
measurements, was suspected on theoretical grounds. Before admitting the hydrogen 
residue to the pipette, the mercury in the dead-space was always lowered to the level 
of the capillary T-piece at which the gas entered. The mercury immediately below 
the dead-space and at the same level in the manometer was cold owing to the brine 
and the ice baths respectively. The result of lowering the pipette reservoir, therefore, 
was to bring cold mercury into the tubing below the dead-space and into the lower 
part of the manometer. The glass in these regions was cooled and if, on raising the 
mercury again, temperature equilibrium was established more rapidly in the manometer 
tube (see p. 414), the points would no longer be set. Moreover, if these assumptions 
are correct, the effect should be greater the higher the temperature of the room. Now 
the room temperatures at the determinations in series 5, ranged from 11°'5 C. to 
13°*5 C., and averaged about 12 C. Examination of the experiments showed no 
connection between the temperature of the room and the value of the result. 
A series of blank experiments was then undertaken when the temperature of the 
room was between 17° C. and 18° C. A setting having been made, the mercury was 
lowered for a period comparable to that required for admitting the hydrogen residue 
in an actual determination. On raising the pipette reservoir again it was found that 
mercury had to be taken from the volume-adjuster in order to set the points. The 
mean volume of mercury withdrawn in four very concordant experiments was 30 c.mm., 
so that actual determinations carried out at this room temperature would have been 
1 part in 20,000 too low. The experiment was repeated at a room temperature of 16" C., 
when the volume of mercury withdrawn was only 10 c.mm. The inference is that, at 
the room temperatures obtaining in series 5, the correction would have been negligibly 
small, and this is confirmed by the above-mentioned fact that differences of 2° C. in 
this temperature region had no detectable influence on the results. It was unfortu¬ 
nately impossible to make blank experiments at these lower temperatures, unless we 
had waited for the winter. 
The error just discussed does not affect the first four series, where, after the hydrogen 
residue had been admitted, and a rough adjustment made by running out mercury, 
the ice in the baths was replenished before the final setting. During this time interval 
the normal temperature conditions of the mercury in the tubes below the dead-space 
level would have been established again. 
For the reasons mentioned in the preface an excess of hydrogen had been taken in 
all our experiments. It does not seem probable that any constant error would be 
involved by this practice, but the point was tested by carrying out a few determinations 
with oxygen in excess. 
