1887 .] Mr John Aitken on Thermometer Screens. 
73 
Even so late as the beginning of November, I was astonished to find 
that the three screens gave very different readings, as will he seen 
from the following maximum temperatures observed on 31st October 
and 1st and 2nd November: — 
Date. 
Stevenson 
Closed. 
Stevenson 
Open. 
Screen 
C. 
October 31 
55*4 
561 
55-0 
November 1 
51-5 
51-8 
50-6 
„ 2 
52'8 
53-0 
52-0 
On these three days the weather was very fine, with sunshine 
and little wind, and on all of these the ordinary Stevenson gave 
readings of from 1° to 1°*2 higher than the C screen. 
These differences in the readings astonished me greatly at the 
time; and when on the 16th of November a difference of more than 
a degree was again recorded in the readings of the Stevenson and 
the C screen, I began to have some doubts as to the correctness 
of the results, as I thought that long before November arrived 
radiation would he so weak that it would not interfere seriously 
with the correctness of the readings given by the Stevenson screen; 
and yet the observations showed that on some days in this month 
the error was very considerable. It was, therefore, thought advisable 
to make some more trials of the screens by means of ordinary ther- 
mometers, and taking readings at short intervals. The morning 
of the 17th November being fine, with little wind and a cloudless 
sky, readings were taken at five-minute intervals. On this occasion 
it was not found possible to make a comparison between the 
ordinary Stevenson screen and the modified form, because, owing 
to the lowness of the sun, the shadows of the tops of one or two 
distant trees passed across the screens, sometimes one and some- 
times the other being in the shadow ; and as the screen in the 
shade always read lowest, it was impossible to compare the open 
with the closed one. They were therefore both worked open, and 
the readings taken of the one in sunshine. These readings, with 
those given by screen C, are shown in the middle series of observa- 
tions in PI. IY. The readings were begun at 11.15 a.m., but the 
Stevenson screens not being dry, no readings were taken with them 
