272 
and the results have been substantially confirmed by 
more direct measurements, namely, by obtaining the 
solar spectrum at different altitudes, in and above the 
maximum ozone level, using instruments carried in 
balloons or rockets. 
The height of the auroral emission has been much 
studied, chiefly by Stérmer; and the spectrum has 
shown the presence of atomic oxygen and molecular 
nitrogen (neutral and ionized) between heights of 100 
and 1000 km, the greater heights being associated with 
sunlit aurorae. The determination of the levels of emis- 
sion of the various spectral components of night-sky 
light is much more difficult, because of the faintness 
of the light. The level of the sodium emission is esti- 
mated as mainly between 70 and 100 km, though part 
of the emission has been acribed to greater heights. 
The red oxygen light is ascribed to a level extending 
upwards from 100 km. A. and E. Vassy conclude that 
there is a second layer of emission at 800 to 1000 km. 
The green oxygen light is estimated to proceed partly 
from 75 to 100 km, and partly from a much higher 
level. These results must be considered as still provi- 
sional. 
21. The Amounts of the Absorbing and Emitting 
Constituents. The amount of a gas which contributes 
to the terrestrial part of the absorption spectrum of 
sunlight can be estimated therefrom in cases in which 
the absorption coefficient is known through laboratory 
or other studies. It is in this way that the number of 
atmo-millimetres of ozone in the atmosphere is deter- 
mined daily at several places. 
In the case of emission it is possible to infer from the 
intensity the number of quanta received per second 
from within a vertical column of atmosphere of, say, 
an area of one square centimetre at the ground. This 
was first done by Rayleigh for the light of the green 
oxygen line in the night sky. He found that 2 x 108 
photons of this light are radiated per square centimetre 
per second, and later studies have confirmed this. The 
sorresponding number for the D line of Na is 8 X 107. 
There is a very strong infrared emission band of OH at 
about 10,400 A, for which the number of photons may 
be 10” or even more. Studies of this kind are still in 
their infancy: they do not indicate the total amount of 
the emitting gas, but they give very valuable informa- 
tion to be fitted into our total conception of the state 
and phenomena of the upper atmosphere [24]. 
As an example of the type of argument that can be 
based on such data, consider the green oxygen emission; 
as it continues all night long with no very great change 
of intensity, the number of quanta emitted from a 
column one centimetre square per night (about 40,000 
sec) is of order 10® . It is not difficult to show that this 
exceeds the number that might be derived from ion- 
electron recombination in the ionized layers of the 
ionosphere, but that the number of the 0 + O — OQ, 
recombinations is more than adequate. Hence this green 
light may draw its energy from the daily dissociation 
of oxygen by sunlight during the daytime, a reservoir 
of energy steadily drawn upon without apparent serious 
depletion throughout the night. The process suggested 
THE UPPER ATMOSPHERE 
is a three-body collision, in which the third body is the 
oxygen atom which is thereby excited so as to emit the 
green light. 
22. Ozone and Atomic Oxygen [26]. The ozone layer 
is produced by photolysis of O. into atomic oxygen. 
The O, absorption of the dissociating radiation occurs 
at different wave lengths and extends from more than 
100 km down to 20 km or so. Let Q: and Q; denote the 
respective numbers of O. and O3 molecules dissociated 
per cubic centimetre per second. Let n, m, ms, 13 
denote the number per cubic centimetre of air mole- 
cules in general, and of O, Oz, and O3 respectively, per 
cubic centimetre. The ozone is formed by attachment 
(in three-body collisions O2 + O + X— O; + X), and 
is destroyed by the reactions 20; — 302 (probably 
negligible in the atmosphere) and O + O; — 20, as 
well as by photodissociation. The atomic oxygen is re- 
moved by attachment and also by combination (20 + 
X — O, + X). These various processes lead to the 
equations 
dny/ dt 2Q2 + Q3 — kynynen — kygnyng 5 
dn3/dt => kyonane — Q3 7 Iygning p) 
dn»/ dt = —Q: — kyninen + kynin =F 2kygnns , 
where the k’s denote coefficients of combination or 
reaction. One difficulty in using these equations to de- 
termine 7, and n3 as functions of height and time is that 
Q3 is known only when n; is known. 
It is possible to explain the seasonal variation and 
latitudinal distribution of ozone on the basis of these 
formulas, using the known type of seasonal variation 
and latitude distribution of Q. and Q3, depending merely 
on the changing zenith distance of the sun. The con- 
stants k are not directly known, but may be chosen 
(with values reasonable from a photochemical stand- 
point) consistent with the attempted explanation. 
The primary process in ozone formation is O, dissocia- 
tion, with the formation of atomic oxygen. With in- 
creasing height above the region of maximum ozone, 
the rate of attachment of O atoms to O. molecules, 
which will be mainly by three-body collisions, will 
decrease upwards as e"/# (see § 11). As the rate of 
ozone formation thus decreases rapidly upwards, the 
life of the free oxygen atoms will increase upwards. 
By simple quantitative arguments it may be inferred 
that at about 100 or 120 km the value of 3 will have 
decreased to less than 10° (from over 10” at the maxi- 
mum level, about 25 km), whereas n, will be of the 
order 10” , comparable with no; and that not far above 
this level ; will exceed no, the ratio n/n. imereasing 
upwards. 
Thus from the observed presence of the ozone layer, 
and from the theory of the associated reactions, the 
increasingly complete dissociation of oxygen as we as- 
cend to high levels can be inferred independently of, 
though in accordance with, the spectral evidence for 
the presence of atomic oxygen as a permanent con- 
stituent of the upper atmosphere. 
23. Atmospheric Sodium. Our knowledge of the pres- 
ence of atomic sodium in the high atmosphere comes 
solely from the emission spectra, night-sky, twilight- 
