641 
of Edinburgh, Session 1885-86. 
By the method here described we do seem to have a practical way 
of surrounding the thermometer with the air to be tested, and of 
causing the surrounding tube to take up and follow the changes in 
the temperature of the air. By means of muslin, cotton wool, or wire- 
cloth, through which the air is made to circulate, it does seem possible 
to surround the thermometer with a tube at the temperature of the 
air, and to protect it from all radiation from surrounding objects. 
Under the conditions existing in the open air, where the case is 
never very highly heated, the error, when the thermometer is pro- 
tected in this manner, need never amount to more than 0°*01, or less 
than a readable amount on all thermometers in general use. 
But though we can by the above means surround the thermometer 
with air at the correct temperature, and with a tube at the tempera- 
ture of the air, yet our difficulties are far from being at an end. We 
unfortunately cannot get the air into and out of the tube without 
openings, and these openings either allow radiant heat to fall on the 
thermometer, or if we place screens to check the radiation, then 
these screens heat the air in its passage over them ; the air and the 
inside walls of the tube get thereby heated, and the thermometer 
gives too high a reading. 
The investigation has, therefore, been continued in this direction, 
and some preliminary experiments made on different methods of 
checking the entrance of radiant heat ; it is unnecessary, however, 
to give here more than a brief outline of the results. For making 
experiments on this radiation effect, the lower end of the draught 
tube was placed horizontally with its opening in front of a large 
gas flame, which was allowed to radiate freely to the thermo- 
meter, the upper part of the draught tube being, as before, vertical, 
to produce the required draught. With the draught produced by a 
small jet of gas, the error due to heating by radiation from the flame 
was 0 o, 45. Screens were then placed inside the case, like louvre 
boards, to prevent the radiation falling on the bulb, but their 
presence was found to increase the error. The effect of a strong 
draught of air produced by a fan was then tried, the screens being 
removed and the flame allowed to radiate direct to the bulb, but the 
effect of the powerful blast was only to reduce the error from 0 o, 45 
to 0 o, 35 ; that is, the strong draught only gave a reduction of T ^th 
of a degree more than the slow draught of the gas jet. This inability 
2 u 
VOL. XIII. 
