11 
1919-20.] The Cooling of the Soil at Night. 
and the rate of radiation was calculated from the value of the rela- 
tive humidity, I suggested that it might be possible to forecast the 
minimum soil temperature for a calm clear night as early as the previous 
afternoon. 
My observations have now extended over a sufficient period to enable 
me to give a formula for forecasting the minimum surface-soil temperature 
and to compare the minima, so forecasted, with the observed minima on 
a number of calm clear nights. 
The observations given in my previous paper were made during the 
winter of 1918-19, and I assumed that the conductivity of the soil was 
uniformly of the value K^’004. This, of course, could only be justified 
if the soil was constantly wet after frequent rain — a state of affairs 
that actually did exist, as can be seen from the following extracts from 
the Monthly Weather Report of the Meteorological Office: — 
November 1918 . . .14 days of precipitation. 
December 1918 . . . 21 
January 1919 . .24 „ ,, 
February 1919 . . . 16 
March 1919 . . . 22 
Thus during the whole period of observation the soil never had an 
opportunity of drying, and the value of the conductivity on any day 
was probably very close to the assigned average value. 
The advent of fine weather in April and May 1919 made it impossible 
to rely any longer on an average value for the conductivity, as it was 
at once apparent, as soon as continuous records were kept of the readings 
of the electrical resistance thermometers, that the conductivity of the 
layer of soil between the 4-in. depth and the surface varied from day 
to day with the degree of wetness of the soil. As soon as a mulch of 
dry soil had formed on the surface, the conductivity diminished rapidly, 
and it appeared that the depth of this mulch determined the con- 
ductivity of the layer of soil down to the 4-in. depth. 
I was thus dealing with a layer of soil which was not uniformly wet 
from top to bottom, and the values of the conductivity obtained were 
more apparent than real ; to avoid this ambiguity, I preferred to employ 
the ratio of the ranges of temperature at the 4-in. depth and at the 
surface (^r ^ as m y standard rather than the conductivity — the more 
so as this ratio is easily observed with considerable accuracy, and is an 
essential part of my forecast equation. 
