RADIATION IN ABSOLUTE MEASURE. 
595 
metal, by first silvering and subsequently depositing copper on the silver, would be 
satisfactory; but finally determined, in spite of the difficulty of joining a metal 
enclosure to the Sprengel pump, to use a hollow copper sphere. 
At the centre of the hollow sphere, the heated globe is suspended and allowed to 
cool, the enclosure being kept immersed in water at a known temperature. The 
temperature of the cooling globe is read off at equal intervals of time by means of a 
thermo-electric junction, and from these readings the rate of cooling, and the absolute 
loss of heat per unit of cooling surface, per unit difference of temperatures of surface 
and surroundings, per unit of time, are calculated. 
The details of the arrangement will be understood from the following description, 
by reference to figs. 1 and 2, Plate 17. In fig. 1, aa is the spherical enclosure. It is 
a copper spherical shell, 10 centims. in internal diameter, constructed like the 
Magdebourg hemispheres, but surmounted by a brass tube, tt, which is brazed to the 
upper opening of the copper shell. At the top of the brass tube is a groove, made 
by brazing on a small piece of wider brass tube ; and into this groove the extremity 
of the glass tube, gg, is inserted. The extremity of the glass tube is prepared by 
grinding to fit the socket; and it is joined in, air-tight, at the last moment, when all 
the other preparations are complete, by means of siegelwaclis, a soft German cement 
which I have faund invaluable for these purposes, and have already mentioned in my 
former paper.By means of a narrow glass tube, ee, the enclosure is attached to the 
Sprengel pump, being hermetically sealed to it in the usual way. An outer tube of 
brass, TT, forms a water-jacket to the tube, tt ; and, by means of a suitably-arranged 
tube, water is kept flowing round the siegelwaclis joint. The copper sphere is sus¬ 
pended in the enclosure by one of the wires of the thermo-junction. 
For the thermo-junction I employ platinoid and platinum. The latter is used 
because it is to be sealed into the glass tube gg, as is shown at s ; and no wire is 
known at present, except platinum, or an alloy of platinum, which, when sealed into 
glass, will make a joint sufficiently tight for a Sprengel vacuum. I have found 
platinum and platinoid to give a very satisfactory thermo-junction; and, along with 
Mr. A. Tanakadate, have published a paper on the thermo-electric position of 
platinoid alloy. + 
The platinoid wire is passed through the brass tube tt at p, and soldered in 
its place. It is, therefore, not insulated ; and it bears the weight of the copper 
sphere. The platinum wire bears no weight. It is insulated all the way up through 
the brass tube by being encased in a very thin tube or sleeve of glass, which is slipped 
over it. The platinum wire has a small length of spiral at the top to take up any 
slack that may exist (as it is not desirable, for glass-blowing reasons, to cut off the 
platinum wire to the exact length), and to avoid any pull on the glass at the place 
where the wire is sealed into the wall of the tube. The platinoid v.ure is cut by 
trial to the proper length, so that, when the weight of the copper globe is on it, the 
* ‘ Phil. Trans.,’ A., 1887, p. 4.3.3. t ‘ Roy. Soc. Proc.,’ vol. 46, 1889, p. 286. 
4 G 2 
