of Edinh argil, Session 1883-84. 
693 
which will heat it twice as much as clean glass when exposed to 
radiation. Any other relation in the heating ehect would do, but 
twice is the simplest to work with. Black paint is too powerful ; 
but I find that coating one half of the bulb with black paint or 
black varnish works very satisfactorily. The advantage of black 
varnish is that it is fairly permanent, adheres firmly to glass, and 
can be easily renewed if required. The black surface ought not to 
be put all on the same side of the bulb, but should be in at least 
two sections, opposite each other, as the radiation from all directions 
may not be equal. 
I have found the working of these differential radiation thermo- 
meters very satisfactory, and it is easily done. Suppose, for instance, 
the blackened bulb reads 69° and the clean one 68°, then the 
temperature 67° is easily found. With practice the eye becomes 
quickly accustomed to the working of the instruments, and easily 
gets the true tenq^erature mentally, even -when dealing with frac- 
tions. These remarks are true only if the thermometers are correctly 
graduated. If the thermometers are not correct, most people will 
require to note down the readings, and make the necessary correc- 
tions, before adding or subtracting the difference. One point of 
importance is, to be very particular about the reading of the lowest 
or clean bulb temperature, in the day observations, as any error in 
it is doubled in the final result. For night observations, it is the 
error in the maximum reading that is doubled, and in this case also 
it is the readings in the clean bulb that have to be most carefully 
attended to. 
For experimenting with these differential readings, I selected 
two thermometers which were nearly correct at the part of their 
scale corresponding to the temperature at the time of the observa- 
tions. These differential radiation thermometers were placed under 
the same sunshade as the thermometer with the silvered bulb. 
This enabled the comparisons to be quickly made, and prevented the 
constantly changing temperature of the air from interfering with the 
results. By watching the differential thermometers till they were 
nearly steady, then rapidly subtracting the difference in their read- 
ings from the lowest or clean bulb, the result generally agreed 
perfectly with the readings of the silvered thermometer, and never 
differed by more than a small fraction of a degree. 
