RADIATION IN ABSOLUTE MEASURE AT VERY LOW TEMPERATURES. 357 
air, are read at intervals and noted alongside the readings with which they correspond. 
Also, the observations, as they are made, are plotted on squared paper. The cooling 
curves thus obtained afford a good check as to the working of the thermo- 
junctions, while they keep the observer informed as to the steadiness of the cooling 
process. 
19. To allow for evaporation of the liquid air, and to keep the level in the Dewar 
vessel up to a fixed mark on the enclosure, a small quantity of liquid air is added 
every 5 minutes. The temperature of the enclosure is kept constant by this method. 
It was found that adding fresh liquid air in large quantities caused irregularities on 
the cooling curves, which were observable as tbe readings were plotted. 
20. Trouble was sometimes caused by the leaking of the copper shell, which, owing 
to the extreme temperature to which it was cooled, would get strained, and allow a 
slight leakage to take place at the joints or the brazed seams. These leakages 
disappeared when the liquid air was removed. They were, however, often overcome, 
and the vacuum kept up to a constant pressure by the Sprengel pumps and the 
charcoal bulb. Latterly the leakage was got rid of by keeping a layer of pentane 
over the fusible metal covering the joint between the flanges, and by painting the 
outside of the enclosure with collodion varnish. 
Calculation of Results. 
21. We will now explain the method of calculating the results. For this purpose 
we use an equation of the form 
_ c | = eS („_V). 
where v is the temperature of the cooling globe, Y that of the enclosure, c the 
capacity for heat of the globe,, and S its surface, while e is a coefficient which has 
been called the Emissivity. It is easy to show by an application of Fourier's 
equations that, to a first approximation sufficient for our present purpose, the loss of 
heat by conduction by the thermojunction wires may be allowed for by adding to S, 
or rather eS, a small quantity depending on the diameter, conductivity, and emissivity 
of the wires, the length of the wires being so great that no heat passes away by 
conduction at the ends. 
22. A point requiring consideration is the following :—The arrangements explained 
above give approximately the temperature at or near the centres of the copper globes. 
In a criticism of D. Macfarlane’s experiments, which were carried out in 1872 with 
the same globes which we have used and under circumstances precisely similar, 
M. Cornu, ‘ Journal de Physique,’ December, 1873, raised an objection to the thermo¬ 
electric method as compared to the method used by Dulong and Petit, where the 
bulb of a thermometer was a cooling body, on the ground that the temperature at the 
