RADIATION IN ABSOLUTE MEASURE. 
437 
probable from a recent investigation by Bunsen,* and from experiments which [ have 
myself made, that when the moisture has been thoroughly removed there is no 
considerable condensation of air at the surface of the glass of the kind frequently 
supposed to disturb the indications of the air-thermometer. It is during the last of 
the heatings just described that the capillary tube be is turned up as shown in the 
figure. When the glass has thoroughly cooled, with the stream of pure dry air still 
flowing through it, a tap leading to the aspirator is closed, and the ends c and d are 
drawn out and sealed with the help of a portable blowpipe, care being taken to 
draw off small portions of the stems in such a way as not to allow any air from 
the blowpipe flame to enter the thermometer. 
I may here, for convenience, describe the subsequent use of the thermometer. The 
bulb is passed into the coil of platinum wire whose temperature is to be determined, 
and the point d is passed through a hole in the asbestos stopper and secured there by 
a small brass wedge. A wrapping of thin asbestos paper is then secured with 
asbestos yarn round the platinum wire and thermometer bulb, and the whole is passed 
into the copper jacket as shown figs. 4a, 4b. A screen of thick asbestos mill-board 
prevents the gases from the heating flame from playing round the point of the 
thermometer. As soon as the heating has commenced the point d is opened with a 
sharp file : and when the temperature to be measured has been reached this point is 
dosed up again by means of a portable blowpipe, care being taken, as before, to avoid 
the introduction into the thermometer of gases from the blowpipe flame. The 
barometric height at the time of sealing is also noted. 
When the jacket and thermometer have cooled down, the latter is removed and 
weighed as follows. First the weight of the glass itself is taken, the air within being 
neglected. Next the point c of the thermometer is cut off under mercury or water, 
and the glass and its contents are weighed. Lastly, the point d is cut off; the glass 
is drawn quite full of mercury or water, and again weighed. The barometric height is 
noted at the time of opening the thermometer, and also the temperature of the 
mercury or water. The expansion of the glass envelope is approximately known ; 
and, if water be used for filling the thermometer, the barometric height is corrected for 
the pressure of water-vapour. The temperature to which the thermometer was raised 
at sealing can be calculated from these data by well-known formulas. 
When water is used for filling the thermometer the precautions pointed out by 
Bunsen and others in connection with gas analysis for avoiding absorption of the air 
by the water are, of course, attended to ; and I find water preferable to mercury for the 
purpose in hand, owing to the largeness of the error introduced by any slight inaccuracy 
as to equality of mercury levels inside and outside the vessel at the first opening of the 
thermometers. 
Thermometers on the same principle as is described here were, I find, used by 
* ‘Wiedemann, Anna 1 en,’ vol. 29, 1886, p. 161. 
