PRESENT STATE OF OUR KNOWLEDGE OF THE UPPER ATMOSPHERE. 93 
series was considerably less than that for other seasons, the actual mean 
annual gradient in the lower layer is less than that deduced from these 
results. The values of the gradient for the first two layers when cases 
of inversion are excluded are 6°4, 5:4, respectively. 
The following values have been deduced from the later manned balloon 
observations, 1901-07 :— 
Height . , . . O-L 1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 km. 
Gradient . 3 , ool bl ors) oe, OO. Tp) Ore ot Ore 
Number of Cases . PDO. BO) 446" 40. 934" ©2210) 83 1 1 
Probable Errorin Gradient— — 02 02 02°02 04 06 — — 
The feature to which Berson drew particular attention was the com- 
parative constancy of the gradient up to a height of 4 km. and the very 
considerable increase in its value in the next layer. The more recent 
observations do not show the peculiarity so markedly and indicate a 
lower level for the discontinuity. Berson attributed the change to the 
fact that the upper limit of the lower clouds is nearly at 4 km. altitude, and 
near this height inversions are more frequent than in the layer above and 
below it. From actual observations in the clouds themselves he deduced 
that the gradient there agreed remarkably well with the theoretical gradient 
for saturated air rising adiabatically, which we may call g,. Just beneath 
the upper limit of the cloud an increase in the gradient was usually observed, 
and just above the upper limit the gradient vanished and the air imme- 
diately above the cloud was generally found to be warmer than that 
beneath its upper surface. It may be noted that the value of g, between 
5 and 7 km. is approximately 7° C. per km., agreeing closely with the value 
found for this region. The mean values for the gradient for each 500 m. 
up to 3,000 m., deduced from the monthly mean temperatures found from 
‘the kite and kite-balloon ascents made at Berlin and Lindenberg, 1903-07,! 
are as follows :— 
Height . 3 . 0-05 0°5-1°0 1:0-1°5 1°5-2°0 2°0-2°5 2°5-3°0 
Gradient . . ey oro 4°6 4-4 48 4:0 5-0 
These values differ considerably from the corresponding values for the 
manned balloon ascents. This may be due to the fact that the kite ascents 
are distributed throughout the year, and are made under a greater variety 
of weather conditions. The large surface value is to be attributed partly 
to the fact that most of the ascents are made between 8 and 10 a.m., and 
the temperature gradient to 500 m. at that time is above the mean 
temperature gradient for the day. 
. Gold? showed that the gradient up to 2 km. depended very considerably 
on the wind direction as well as on the time of the year. He found that 
inversions were most frequent in winter and with easterly winds ; that 
they occur very rarely indeed with N.W. winds, and then in summer, 
a season when they are not found with winds from other directions. 
Field * made kite ascents in India and over the Arabian Sea during the 
S.W. monsoon, and found a very rapid decrease of temperature up to 
300-400 m. At greater heights up to 3,000 m. the gradient was very close 
to that for saturated air rising adiabatically, 2.e., about 5° C. per km. 
Hann ‘ deduced from mountain observations that the mean tempera- 
ture gradient up to 3 km. is 5°7 to 5°8 per km. The earlier balloon 
1 Ergebnisse Aeronautischen Observatoriums, Berlin and Lindenberg. 
2 Barometric Gradient and Wind Force, M.O., No. 190. 
* Indian Met. Memoirs, vol. xx. part 7. 4 Lehrbuch der Meteorologie, p. 104. 
