174 Proceedings of the Koyal Society of Edinburgh. [Sess. 
amount of air. If the thermal inertia is very large, the thermometer will 
give roughly the average temperature of a greater amount of air than in 
the former case. Large bulbs and certain forms of screen occasion a 
further source of error on account of radiation effects. 
With these considerations as a guide, two exactly similar* Fahrenheit 
thermometers with bulbs 10 mm. in diameter were used. The scales were 
wide and easily read, and the Kew corrections were very small, never more 
than 0°T at any part of the scale. They were placed under two sunshades 
of my own design, of which one variety is shown in the accompanying 
figure (p. 176). 
This screen consists essentially of two horizontal boards one above 
the other, with an air space between, as indicated in the sketch. The 
thermometer bulb lies a short distance below the lower board, and the 
stem passes upwards through holes pierced in the boards. This form of 
screen is more efficient than the Stevenson arrangement. A thermometer 
protected by it always reads lower than the thermometer in the Stevenson 
screen when there is any radiation, although the latter may read as low as 
the former at the beginning of the rise in the presence of sunshine. This 
is owing to the great thermal inertia of the Stevenson screen. Further,, 
the errors in the readings of the Stevenson screen thermometer go on 
increasing with the duration of the radiation. There is no such effect in 
the case of the other form of screen, and all maximum readings taken 
under it are lower than those taken in the Stevenson screen. 
There are several reasons for the temperatures within the Stevenson 
screen being recorded too high. First, the air comes into the screened 
region over louvres heated by radiation. Second, the open bottom of the 
screen allows radiant heat from the soil to enter. Third, the thermometer 
readings are also raised by the heating of the inside of the roof during 
continuous radiation. This may be easily remedied by fitting to the screen 
another cover, with an air space between. 
We return to the experiments with the two thermometers. These 
were placed with their bulbs about 4 cms. below the underside of the 
screen. The readings were always the same even under the condition of 
strong sunshine. 
The bulb of one of the thermometers was now coated with aluminium 
paint, which is known to have small absorption powers for heat, and 
placed under the sunshade alongside the other thermometer. The readings 
of the two remained practically the same. Gold paint was then tried, with 
the result that the reading of the thermometer with the gilded bulb fell to 
nearly one degree below the other, showing therefore considerably less 
