256 
As is to be expected, auroral activity and magnetic 
disturbances are closely associated with solar phe- 
nomena. For instance, both auroral and magnetic activ- 
ity show remarkable correlation with sunspot numbers 
[28]. Also both have seasonal variation—maxima in 
March and October and minima in June and January 
[27]. 
Weather and the Upper Atmosphere. It may seem 
surprising that weather conditions in the tropospheric 
regions should have any association with the physical 
conditions of the atmosphere tens or even hundreds of 
kilometres above the surface of the earth. Nevertheless, 
evidence of such associations (some of them yet un- 
explained) has been obtained. 
It has been found that the ozone content in the 
middle atmosphere is markedly related to weather con- 
ditions in the troposphere [31, 32]. Almost without 
exception the ozone value is found to be high with 
cyclonic systems (low pressure) and low with anti- 
cyclonic systems (high pressure). Periods of fine and 
settled weather have been observed to be associated 
with relatively low ozone value [101]. 
Atmospheric densities at heights of 75 km (as deduced 
from observations on meteors) show a marked seasonal 
variation, and this variation is correlated with the mean 
ground temperature [105], a maximum density being 
indicated at the time of maximum ground temperature. 
Association between Region E ionization and thun- 
derstorms has been reported by some observers [6, 17, 
92, 93]. Evidence has also been obtained of association 
of weather conditions in the tropospheric regions with 
variations of ionization densities in the ionospheric 
regions [60]. For instance, it has been observed that the 
variations of the minimum heights of Region F and of 
the Region E tend to follow the variations of the 
barometric height. Again, day-to-day variations of noon 
ionization of Region E seem to correspond to the day- 
to-day variations of barometric height. Some remark- 
able correlations of the variation of Region F2 ioniza- 
tion and cyclonic conditions near the ground have 
also been reported from Australia [9]. Other similar 
correlations, scarcely less remarkable, have been re- 
ported from Shanghai, China [38]. It has been found 
that there is close relationship between the occurrences 
of the E, F, and F»2 echoes (when the ionosphere appa- 
ratus is set to the fixed frequency, 6 me sec~!, the mean 
E-critical frequency of the place) and the movements 
of the three main air masses—polar, maritime, and 
equatorial—which cause weather all over the world. It 
is claimed that these observations can be very success- 
fully used for weather prediction. 
Extension of the Atmosphere to Great Heights. 
Streamers of aurorae of the ray type extend to great 
heights (800 to 1100 km) into the sunlit portion of the 
upper atmosphere. Spectroscopic study of the auroral 
light shows that even up to such heights the atmosphere 
consists of atomic oxygen and molecular nitrogen. 
SOME UNSOLVED PROBLEMS 
The brief account of the contemporary state of our 
knowledge of the upper atmosphere given above shows 
THE UPPER ATMOSPHERE 
that in spite of the many difficulties, a considerable 
amount of information has already been gained. How- 
ever, there are still gaps in our knowledge. This is due 
not only to the inaccessibility of the regions concerned, 
but also to the lack of many fundamental laboratory 
data of the atmospheric constituents. But perhaps the 
greatest obstacle has been our imperfect knowledge 
regarding the nature of extreme ultraviolet radiation 
from the sun. Let us hope that this gap will soon be 
filled up by observations made with the help of V-2 
rockets. 
Let us attempt a survey of the upper atmospheric 
problems which need further investigation or are still 
awaiting solution. It is very satisfactory that direct 
observations with V-2 rockets have confirmed the sur- 
mise made long ago that the main atmospheric con- 
stituents up to about 70 km occur in the same propor- 
tion as near the ground [23]. It is very probable that 
not far above this level the dissociation of O2 molecules 
begins. But the relative distributions of O. and O with 
height are not known with certainty. Several possible 
distributions have been obtained by different authors 
under different assumptions [25, 57, 88, 90, 108]. But the 
distribution which best represents the actual state of 
affairs has still to be found. Accurate determination of 
this distribution is very necessary because, according 
to more than one author, the electron-density distribu- 
tion in Region E depends upon the distribution of 
molecular oxygen in this transition layer [16, 67]. 
The question at what height above 70 km diffusive 
separation becomes important is of considerable interest 
because on it will depend the relative abundance of N2 
and O in the highest regions of the atmosphere [75]. 
With diffusive separation the highest regions should 
consist almost entirely of O atoms. But, as mentioned 
earlier, the spectrum of auroral streamers up to 1000 
km contains atomic oxygen lines and molecular nitrogen 
bands with comparable intensities. It is still not clear 
how nitrogen molecules which are nearly twice as mas- 
sive as oxygen atoms reach such great heights. 
In the ozonosphere the modes of production and 
destruction of ozone and the ozone equilibrium resulting 
therefrom are in need of further elucidation. Interesting 
relations have been found between the ozone content 
of the atmosphere and weather conditions. But the 
ozone observations are still confined to a few stations. 
It is extremely desirable that a world-wide network of 
stations be established for simultaneous and systematic 
studies of atmospheric ozone, of temperature distribu- 
tion in the middle atmosphere (by abnormal sound 
propagation method), and of meteors (by radar tech- 
nique (34, 46]). Such studies will be helpful m under- 
standing the correlations that have been observed 
between weather conditions in the troposphere and 
temperature variations in the middle atmosphere and 
may also be of help in long-range weather forecasting. 
The temperature distribution in the region 70-90 
km needs more accurate determination. Rocket ob- 
servations confirm the results of studies by various 
indirect methods indicating that there is a rapid drop in 
temperature in this region. But the magnitude of the 
