Thermometer Screens. 
1 77 
1919-20.] 
the rays fall very obliquely on the surface of the radiation box. For no 
particular height of the sun, however, do these two types of instrument agree 
in their maximum readings, since the radiation box is greatly affected by 
the wind. The temperature of the radiation box is also raised by long- wave 
radiations from sky, cloud, and surrounding objects. As already pointed 
out in previous communications, the indications of the radiation box are more 
in accordance with climate than are the readings of the black bulb. 
The accompanying table (p. 180) contains observations from August 17, 
1919, to October 81 of the same year.* A brief discussion of these will 
illustrate the points now being emphasised. 
To all temperature readings the Kew corrections of the thermometers 
have been applied. The observations here recorded were all made at 9.30 a.m. 
The first item entered for each day is the direction and strength of wind. 
Then follow the temperature readings of the thermometers in (1) the 
Stevenson screen ; (2) the screen C as shown in fig. 2, without the annular 
piece x; (3) the difference between these ; (4) the temperature registered by 
the thermometer in screen D, which was provided with a vertical draft tube. 
Then follow the readings of the radiation box, B.B., and the black bulb in 
vacuo , B.B. ; and finally a general description of the weather for the day. 
An inspection of the table will establish clearly the following points : — 
1. The thermometer in the Stevenson screen reads always higher 
than the others. 
2. The error of the Stevenson screen thermometer is greatest when 
the B.B. and B.B. readings are high and there is an absence 
of wind. 
3. The error is small when the weather is dull and the B.B. and 
B.B. readings are slightly above the temperature of the air. 
These features are brought out very strikingly in the week beginning 
September 20. The fine, sunny, comparatively calm weather of the 21st 
and 23rd days are associated with large errors in the Stevenson screen 
thermometer ; while the errors are very small on the contiguous days of 
cloudy and stormy weather. The long stretch of fine weather beginning 
October 6 and ending October 16 is characterised by large errors of the 
Stevenson screen thermometer and high readings of the radiation box. 
[The average results may be brought out very simply by forming the 
data into three groups: (1) those in which the error of the Stevenson 
screen thermometer is 1*5 and upwards; (2) those in which the error lies 
[* In the I original notebook from which these observations are taken, the daily 
observations continue till November 8, six days before Dr Aitken’s death.] 
VOL. XL. 
12 
