Comparison of Formula ''for Total Radiation. 437 



Before beginning any one series of observations, readings 

 were taken from each Of the three thermometers. These had 

 of course been previously compared and found reliable, so 

 that the correction for difference between them was very 

 small. The disk had been hanging in position for some hours, 

 but on exposure the thermopile always indicated a slight dif- 

 ference of temperature between itself and the disk, amount- 

 ing to one or more scale divisions. This deflection was 

 recorded, to be used as a correction to subsequent readings. 

 The disk was then removed, heated up to a given temperature 

 in excess of any to be measured, and hang again in position. 

 After a few minutes its temperature was noted, the thermopile 

 exposed long enough only to produce a throw of the galva- 

 nometer needle, the temperature of the disk immediately 

 again noted, and also the temperature of the thermopile as 

 indicated by the thermometer in the box. This procedure 

 was repeated at intervals of two minutes during one or two 

 hours, in which the disk became cooled down to a temperature 

 differing but little from that of the room. 



The arrangement of apparatus and the procedure just de- 

 scribed was that finally adopted after many days had been 

 spent and many hundreds of observations taken which had 

 failed to yield satisfactory results. With the utmost care it is 

 impossible to exclude air currents from the disk, and hence 

 irregularities in its radiation. The same remark would apply 

 equally to the Leslie cube, with the additional consideration 

 that convection currents within the cube during its cooling 

 would be very hard to control, even with assiduous stirring. 

 At each exposure of the thermopile the air around it becomes 

 warmed slightly by the disk, and it cannot be assumed that the 

 temperature of the room is that of the thermopile after the 

 first few exposures have been made. By enclosing the instru- 

 ment in a box, this warm air, it is true, is prevented from 

 passing away ; but this disadvantage is much more than offset 

 by the possibility of measuring and recording the increase of 

 temperature, for which a curve of corrections is afterward 

 made. 



To continue any one series of observations until the differ- 

 ence of temperature between disk and thermopile vanishes is 

 obviously impossible. Even when this difference amounts to 

 several degrees disturbances are apt to produce such irregu- 

 larities in the indications of the thermopile as to introduce 

 serious error. By artificially cooling the disk until it seemed 

 to indicate the same temperature as the box it was found that 

 uncontrollable errors were introduced. It was best therefore 

 to allow the cooling by radiation to continue until a tempera- 

 ture-difference of 5° or 10° remained, recording the corre- 



