( 804 ) 
somewhat greater than that reached by the column of liquid was 
everywhere the same both in the wide and in the narrow tube; 
and further care was taken that where the temperature of the vapour 
above the liquid in the upper parts of the apparatus changes to 
ordinary temperature it was as far as possible the same at the same 
height in the two limbs of the 0-shaped space. With this end in 
view the liquid in the inner tube was, by means of the magnetic 
field, repeatedly moved up and down under constant vapour pres¬ 
sure, until we might assume that in this tube equilibrium was suffi¬ 
ciently well attained. To make this equilibrium possible the inner 
tube is surrounded with liquid at the same temperature as that 
which the liquid in it must attain. In the outer or protecting 
tube the liquid is kept constantly in motion by means of a stirrer 
consisting of a brass ring S l that can be moved up and down; it is 
possible to do this and still keep the space closed by utilising the 
flexible rubber tube The vaporization in the inner tube is thus 
very small (between 0.5 and 1 litre of gas measured under normal 
atmospheric pressure escapes per hour). 
Notwithstanding all these precautions temperature differences 
must still be encountered. In the liquid, in which the convection 
currents maintaining heat-equilibrium can be followed by the 
small particles which they carry along with them, these tempe¬ 
rature differences must have been very small. In the gas layer 
in the upper portion of the O-shaped space there must indeed have 
been considerable differences; but on account of the small density 
of the gas, these have but small influence upon the difference of 
level in the two limbs, and, moreover, that influence may be almost 
entirely neglected seeing that the observations are simply comparative 
measurements with and without the magnetic field. Now, care has 
been taken that the temperature over the distance that the liquid 
rises can vary but slightly, while in the upper portions of the tube 
practically the same state of affairs is maintained during both obser¬ 
vations. We have therefore omitted the correction that should still 
have to be applied for possible temperature differences. 
Comparing the positions of the liquid in the narrow cylindrical 
tube with and without the magnetic field also reduces the correction 
for capillarity to the insignificant differences in form of the menisci, 
and this correction, too, we have omitted. 
The inner and the outer tubes are closed independently of each 
other by means of the German-silver caps P„ P 4 , Q lt Q 4 (fig. 1); 
the junction is made air-tight by the rubber sleeves M 0 , JST,, which 
at the same time unite the two tubes firmly together. Liquid oxygen 
