1920-21.] Evaporation of Liquid Air in Vacuum Flasks. 109 
reached the neck in its upward path. With these flasks the gradient near 
the base of the neck, though slow, was not zero ; a certain amount of heat, 
therefore, reached the inner sphere by the neck. The graph indicates, 
however, that the transference of heat due to this cause was here equivalent 
to that which would have been yielded by a neck of the same sectional 
dimensions about 64 ins. long in which the gradient was uniform from 
top to bottom. 
The fourth container, as already stated, was a very large German vessel ; 
at the time of testing, it was found to be discharging gas at the rate of 3 *3 
litres per minute, measured at 18° C. It was constructed of brass, with a 
German-silver neck 24f ins. long and xV-in. bore. The thickness of the 
metal of the neck could not be determined ; the plug (E, fig. 1 ) probably 
extended to a depth of about 3-J ins. from the mouth of the bottle. The 
measurements (see fig. 3, curve marked “ German 300-lbs.”) show that the 
passage of heat to the liquid via the neck was zero, and that the rate of 
evaporation would not have been affected in the least had the neck been 
half the length. 
In general, the results indicate that container necks may be con- 
siderably shortened, or both shortened and thickened, without appreciably 
increasing the rate of loss of the flasks. Such an alteration in construction 
will strengthen the flask and save weight, and, while preserving sufficient 
flexibility in the neck to allow of the spheres touching during the act of 
pouring, the excess loss during transportation — which is principally due 
to the continual bumping together of the cold and hot globes — will 
be lessened. Underground transport (a matter that the present writer 
has especially in view) will be facilitated by the reduced height of 
the bottle. 
Results obtained with Short-necked Containers. 
Believing at that time that container necks could be shortened and 
strengthened without any serious effect upon the evaporation rate, the 
writer designed, in 1918, a 50-lb. container of which twelve were made and 
ten proved sound. Though there were in these bottles a number of 
variations upon the standard design illustrated by fig. 1, only two of 
them could have any influence upon the rate of evaporation. The first of 
these — namely, the provision of a loose, insulated cap to fit over the mouth 
of the bottle — was found to have only a slight effect ; the cap, when in place, 
only reduced the loss by a few ounces per diem, a fact which itself 
demonstrated the relatively small heat-inflow by the neck. The second 
variation was in the size of the neck, which in each of these special 
