of Edinburgh, Session 1885-86. 
637 
rounded by the case beated to 100°; in column 3 the difference 
between the thermometers, and shows the amount of heating pro- 
duced by the warm case ; and in column 4 the protection placed 
between the thermometer and the hot case. 
It will be observed that when there was no protection between 
the thermometer and the case, the thermometer got heated 1 8° *2 
above the temperature of the air, or three times more than it 
was before the radiating surface was painted black. It will be 
noticed that a single thickness of muslin placed between the ther- 
mometer and the case greatly checked the radiation ; that as the 
number of thicknesses was increased, the heating effect rapidly 
diminished ; and when twelve folds or thicknesses were between 
the two, not the slightest effect could be detected. This test 
has been repeated a number of times with the same result. To 
check the results, readings were taken with the steam off and the 
case cold. Steam was then turned on, the case heated, and readings 
taken. Steam then cut off, and cold water circulated through the 
jacket, but not the slightest difference could be detected under the 
two conditions, showing that the thermometer was thoroughly pro- 
tected by twelve thicknesses of muslin. The protection was of 
course perfect only under the condition of the experiment. If the 
draught was reduced by shutting off the gas, and the air heated only 
by the small amount it got in its passage through the case, then the 
thermometer rose a very little above the temperature of the air. 
No comparative trials were made with black and white muslin, 
as it was not possible to get two muslins similar in every respect 
except colour; but as we are here dealing with dark heat, it does not 
seem likely that this would have any effect. 
The next two tables give the results of tests made with two 
different kinds of wire-cloth as the protecting material. In the 
first table are given the results obtained with wire-cloth made with 
No. 35 iron wire and 36 meshes to the inch; in the second table 
the results obtained with No. 38 brass wire and 70 meshes to the 
inch. 
The contents of the different columns in these two tables are 
arranged as in the last one. It will be observed that the results 
with wire-cloth are not so good as with the muslin; this will be 
due partly to the conducting power of the wire, and also probably 
