316 THE EARL OF BERKELEY, MESSRS. E. G. J. HARTLEY AND C. Y. BURTON: 
air passing through the whole apparatus is approximately proportional to the area 
of the filter not covered by mercury, and it may be stated that we have found this 
device most useful, especially when it is provided with a graduated scale to enable 
one to set the mercury level to a definite height. 
Before reaching C, the air has passed a side-tube which connects with the vessel D. 
This vessel (of 5 litres capacity), containing a small quantity of water, the level of 
which is observed (telescopically) in the graduated capillary tube D 2 , serves three 
purposes. With the tap E closed, and the mercury in C adjusted to the proper 
level, the rise of the water in D 2 is a measure of the rate at which the air is passing ; 
thus any obstruction in the air stream will at once be indicated, while the amplitude 
of the oscillations (if any) of the water meniscus affords a means of detecting outside 
atmospheric pulses. Again, with the tap E open (and this is the normal position 
during an experiment) the outside air pulses are practically eliminated, and any 
oscillations that may then be apparent are the result of changes in pressure on the 
exhaust side. It was found that when the iron tank G was added to the apparatus 
the pulses were so reduced as to be scarcely perceptible. 
The air normally passes through tap F into G (200 litres capacity), and tap H is 
kept closed ; this latter tap is only used in conjunction with vessel J when testing 
the various joints for leaks. A partial vacuum is maintained both in G and in the 
15-litre jar K by a Fleuss pump (not shown) which is joined on at L. In the main, 
the degree of vacuum is determined by the height of oil in K, above the air inlet M. 
We will now consider the various corrections that have to be applied to the 
experimentally determined vapour pressures. 
Burton’s Correction .—-The most important of these is that which we will call 
“ Burton’s correction,” for it was he who showed that the effect is not negligible as 
we had erroneously assumed. The correction is inherent in the method itself, and 
will be apparent at once when it is realized that the air stream (which has been 
saturated up to the vapour-pressure of the solution) on entering the water vessel is 
expanded slightly by the vapour it takes up there ; consequently the volume of air 
when leaving the water vessel is slightly greater than when it left the solution. 
This correction, together with the effect due to changes in barometric pressure, was 
briefly outlined in ‘ Nature ’ (March 11, 1915, p. 34) ; there are, however, other factors 
involved which make a more general discussion desirable. 
If we make the assumptions that the temperature of both solution and solvent is 
the same and remains constant, and that the air space between them is negligibly 
small, and further, that the air stream flows slowly enough for complete saturation 
and freedom from turbulent motion, then the following analysis (for which we have 
to thank Mr. G. W. Walker) will be applicable. 
Let p be the pressure of the air at any point in the train of vessels and r its rate of 
passage in cubic centimetres per second. Let ir n and i r 0 be the vapour pressur.es of 
solution and solvent respectively and p x and p 0 the corresponding vapour densities. 
