130 Proceedings of the Royal Society of Edinburgh. [Sess. 
It would appear that, during the period November 1918-March 1919* 
the following points were borne out by actual observations : — 
(1) That the radiation from the soil may be accounted for in counter- 
balancing the upward conduction and the latent heat of freezing — 
the residue only cooling the soil. 
(2) That the rate of radiation of the soil on calm clear nights, when 
5th magnitude stars are visible, is a function of the relative 
humidity. 
(3) That other causes such as condensation, evaporation, etc., have little 
effect on the temperature of the soil on calm clear nights. 
(4) That the surface tends to fall rapidly such a number of degrees 
below the temperature of the 4 in. depth as will make the conduc- 
tion from this depth balance the radiation ; after this takes place 
the surface temperature can fall no faster than that of the 4 in. 
depth, and a sufficiently high temperature underground will 
obviously render a frost unlikely. 
(5) That this temperature difference between the surface and the 4 in. 
depth, which makes the upward conduction balance the radiation, 
is probably about 10° F. during the winter, when the soil is 
almost invariably wet, and of uniform maximum conductivity 
from day to day, but may be as much as 20° F. after a dry spell 
in spring or early summer. 
(6) That the prediction of frost on any given night depends on the 
possibility of assessing the value of the following : — 
(a) Average relative humidity during the night. 
( b ) The temperature of an assigned depth— say 4 in. — at the 
time of surface minimum. 
(c) The conductivity of the layer between the assigned depth 
and the surface. 
( d ) The difference between the surface soil minimum and that 
of the air above it. 
(7) That it might be possible, after an extended series of observations 
with a set of electrical resistance thermometers, to forecast the 
minimum temperature on calm clear nights from observations 
taken in the early afternoon. 
With a view to thoroughly investigating this question, I have recently 
had a set of four electrical resistance thermometers installed, and have taken 
hourly readings through the day and on clear nights during the last seven 
weeks (March and April 1919). 
