WATER VAPOUR IN THE UPPER AIR 
lower than 180K. Thus if the thimble should be cooled 
below this temperature, the ice will form as a glassy 
layer which scatters no light. 
In using the eye-observation hygrometer it is essen- 
tial that two thimble temperatures be found, one at 
which the deposit is slowly evaporating and the other 
at which it is growing equally slowly. The mean of the 
temperatures will be the frost point. Since a deposit 
will usually not form on a clean thimble until it is 
cooled to near the dew point, large errors will be made 
if the temperature at which the deposit is first seen is 
used. The temperature at which the deposit finally 
evaporates is also of no significance. With photoelectric 
indication of the amount of deposit, one can adjust 
the thimble temperature until the deposit mdicator is 
steady and then read the thimble temperature, which 
gives the frost point directly. 
Figures 1 and 2 show the general design of the types 
of hygrometer used in Great Britain [4, 6]. 
PHOTOCELL 
LAMP 
TZZZZZZZZLZ LEELA 
THERMO JUNCTION 
THIMBLE 
SOLID COp 
AND PETROL 
PUMP CONTAINING 
COLD PETROL 
Fie. 2.—The frost-point hygrometer with photoelectric de- 
posit indication. A deposit on the top surface of the thimble 
scatters light which is received by the phototube. The thimble 
is at frost point when a small deposit is neither growing nor 
evaporating. (Reproduced by courtesy of the Royal Society, 
London.) 
THE RESULTS OF AIRCRAFT ASCENTS 
The measurements which have so far been made 
show that wide variations of relative humidity occur 
in the atmosphere, and any relative humidity between 
30 and 90 per cent, with respect to ice, is about equally 
common. Relative humidities of about 5 per cent are 
relatively frequent, and the lowest which has yet been 
observed is 0.65 per cent. This humidity was observed 
313 
at an air temperature of 266K and a frost point of 
219K. It is characteristic of the frost-point hygrometer 
that even low humidities of this kind can be measured 
precisely. If the relative humidity were, say, 0.75 per 
cent, this would correspond to an increase in the frost 
point by one degree centigrade to 220K, which could 
be measured quite readily. A similar relative humidity, 
0.65 per cent, has also been observed in the stratosphere 
with an air temperature of 225K and a frost pomt of 
189K, but this could not be measured so precisely. 
Supersaturation with respect to ice is a relatively fre- 
quent occurrence, but supersaturation with respect to 
supercooled water has not been observed. 
Comparison with Radiosonde Measurements. The 
comparison of radiosonde humidity measurements with 
humidity measurements made by a dew-point hygrom- 
eter is difficult because the air is often patchy and any 
differences may not be instrumental unless the ascents 
are made at the same time and place. Two ascents have 
been made by Brewer and Harrison [5] in which the 
aircraft from which the dew-poimt measurements were 
made cireled the ascending radiosonde which had a gold- 
beater’s-skin hygrometer. The two ascents are shown 
plotted as relative humidity against height in Fig. 3. 
> 
ASCENT 4 
NO. | 
> 
KILOMETERS 
~ 
a 
to) 
le} 25 50 75 100 
RELATIVE HUMIDITY (PER CENT) 
Fig. 3—Comparison of humidities as measured by frost- 
point hygrometer (broken curve) and Kew pattern radiosonde 
(solid curve). The aireraft from which the frost-point hygrom- 
eter measurements were made circled the balloon, which 
was operated at reduced lift. (Reproduced by courtesy of the 
Physical Society, London.) 
It will be seen that the radiosonde very greatly smooths 
the curves and in ascent No. 1 does not show the dry 
layer at 2 km where the lowest humidity is 1.0 per cent. 
Below 1 km the difference between the radiosonde and 
the aircraft hygrometer is due to the presence of broken 
cloud layers. The aircraft was flown through the dry 
air between the clouds, but a balloon tends to become 
entrained in clouds. 
The Lower Stratosphere. In the ascents which have 
so far been made into the stratosphere (about seventy, 
all in southern England), the frost point generally falls 
with height in the region of the tropopause, but imme- 
diately above the tropopause there is an increase in the 
rate of fall, and one or two kilometres above the tropo- 
