206 
MR. GLAISHER ON THE RADIATION OF HEAT, 
In this respect the results are somewhat different from those obtained from simul- 
taneous observations of mercurial thermometers ; but it would certainly seem that a 
thermometer placed on a sheet of metal on the ground will read as low as that in the 
reflector, although that of the latter would attain its lowest reading in less time than 
that of the former. 
From the numbers in this table the numbers in the first ten columns of Table CVI. 
are formed. 
As in a previous part of this paper it has been stated that on cloudy nights a 
thermometer read differently according as the cloud was high or low, the following 
Table has been formed by collecting all those cases in which the night has been 
cloudy throughout. 
Table CIII. — Showing the results on totally cloudy nights, but the clouds not 
cirrostratus. 
The sky covered with cloud during the night, the cloud other than cirrostratus. 
Year. 
Month, day 
and hour of 
reading the 
instruments. 
The excess of the reading of a self-registering minimum therm, placed in air at the height of 4feet 
and protected from radiation, above that of a similar thermometer fully exposed to the sky, placed 
On long 
grass. 
On short 
grass. 
On white 
raw wool. 
On fine 
flax. 
On coarse 
flax. 
On 
white un- 
wrought 
cotton 
wool. 
On 
blackened 
tin. 
On white 
tin. 
In focus of 
metallic 
parabolic 
reflector. 
1842. 
d h 
Nov. 7 21 
o 
o 
O 
O 
O 
O 
O 
O 
O 
9*3 
o 
9 21 
9-0 
17 21 
9-4 
20 21 
8-7 
1843. 
Feb. 14 21 
13-3 
25 21 
8-5 
Nov. 10 21 
9*7 
7-6 
12-6 
12-3 
13-4 
12-2 
9-5 
32-6 
1844. 
Feb. 14 21 
8-5 
7-6 
14-8 
13*8 
7-3 
6-8 
7-5 
36-8 
May 9 21 
10-7 
9-9 
9-3 
9-4 
11-4 
51-7 
Oct. 10 21 
10-0 
6-4 
11*5 
7-9 
44-5 
Table CIV. has been formed by collecting all those cases in which the night has 
been cloudy throughout, and the clouds have been noted as being high ; and Table 
CV. has been formed in a similar way from all those marked low. Those cloudy 
nights during which the height of the cloud was not noted have not been used. 
The means of the results in Tables CIII., CIV. and CV. are contained in Table CVI. 
