424 
DR. F. F. BURT AND DR. E. C. EDGAR ON 
constituted the first hydrogen filling of the succeeding experiment.* The results 
show no indication of an accumulating impurity. 
The successful removal of water from the hydrogen residue was proved by the fact 
that the small phosphoric oxide tube between the explosion vessel and the pump showed 
no signs of deliquescence after four years’ use. 
Accuracy of Volume, Pressure, and Temperature Measurements. 
Owing to the fact that all gas measurements were made in the same vessel under 
the same conditions, various possible errors in this category tend to cancel out. 
Volume errors. —Since all volumes were measured in terms of a weight of mercury, 
an accuracy greater than necessary was readily secured. The only operation in which 
special care was needed to prevent a volume error was the setting of the meniscus 
to the point in the dead-space. The final adjustment was always made with a 
rising meniscus, and cohesion effects were minimised by tapping the glass or pinching 
the reservoir tubing. The probable limits of variation may be estimated from the 
concordance of the values obtained in calibrating the dead-space at the beginning of 
the research. 
Pressure errors. —The vertical distance between the two glass points of the 
manometer was measured with a cathetometer when the system was in ice. The 
accuracy, attainable was not very great because of distortion by the small bath round 
the dead-space, but from two measurements giving values 760'35 and 760‘32 mm., 
respectively, it may be concluded that the distance was within a third of a millimetre 
of the normal barometric height. A deviation of 1 mm. alters the density ratio only 
by 1 part in 500,000, so that a difference of the above order is quite negligible. 
The precision with which the mercury meniscuses could be set to the two points 
was certainly very high. The order of accuracy attainable was tested in the 
following way : the upper chamber of the manometer was viewed through a telescope 
fitted with a Hilger micrometer eye-piece, and the cross-wire was set on the point. 
The meniscus was then adjusted until the cross-wire appeared exactly tangential to 
it. On examination with the pocket lens which was used in making the settings, a 
distinct gap was detectable between the point and its mirror image. Assuming the 
limit of accuracy with a micrometer to be 0‘01 mm., it was estimated that adjustment 
by means of the lens should be within 0‘002 mm. 
The effect of capillarity on the pressure measurement can be neglected, since in 
tubes of the diameter of the point chambers (16 mm.), even the absolute depressions 
are vanishingly small. 
Temperature errors. —These were undoubtedly the limiting factors in the 
accuracy of the work. Several have already been discussed in detail in the course 
* At this stage of the research the bubble trapped in the dead-space was removed after expulsion of 
each pipette-full of gas, so that the final hydrogen filling was not contaminated with oxygen. 
