1887.] Mr John Aitken on Thermometer Screens . 83 
with each other and with the silvered bulb, at the more exposed 
Granton site than at the more confined one here. Further, on both 
days on which the Granton trials were made, the wind entered 
the screens from the cold side. This of itself is a most important 
point, especially if the sky is clear, as the cold sides of the screens 
may be below the temperature of the air, and the passage of the 
air in that direction prevents the advance inwards of the heat from 
the sun-lieated louvres. I have noticed in the trials here that on 
all occasions on which the wind was north of east or west the errors 
were comparatively small. It therefore seems possible that the more 
open exposure of the Granton site, together with the direction of 
the wind at the time the two sets of the Granton observations were 
made, are the principal causes of the difference in our results. 
Since this paper was given in, Mr Dickson, who has continued 
his trials with the Stevenson screens at the top of Ben Nevis, has 
very kindly furnished me with an abstract of the result obtained in 
that situation. He says the readings for thirty-four days gave the 
following mean maxima : — 
Stevenson, Open bottom, 41° - 45. 
„ Closed „ 41 c, 00. 
Thus showing a difference in favour of the closed bottom of nearly 
half a degree, a result confirming the conclusion arrived at in this 
and a preceding paper. 
The site here being surrounded by trees in almost every direction, 
may in part explain the reason why screen C, even when the 
annular piece F was removed, gave such correct readings. We have 
seen that on the 7th October the temperature of the grass rose 18° 
above the temperature of the air. Now that is the temperature to 
which the thermometers in the Stevenson screen with open bottom 
are exposed, while screen C, by its construction, cuts off all this 
radiation and exposes the bulb to the radiation from the trees, which 
will never be so highly heated as the grass, as they have a freer 
circulation of air through them. This, combined with the very 
free circulation of air through the screen, and the method adopted 
for preventing the heated air touching the thermometer, would seem 
to account for the low readings given by this screen even when the 
piece F is removed. 
