314 
pause the air normally has a frost point of about 190- 
200K and a humidity with respect to ice of 2 or 3 per 
cent. The effect is normally striking and consistent. An 
ascent is shown in Fig. 5, and a mean curve (due to 
Shellard [15]) of the results so far available (southern 
England) is shown in Fig. 4. 
MILLIBARS 
+100 \ \ 
[ *‘FROST- 
POINT 
Fic. 4.—Mean variation of temperature and frost point with 
height above and below the tropopause, for southern Hngland. 
All data so far available. (Curve due to H. C. Shellard (15].) 
This effect had been predicted by Jaumotte [11], but 
other workers on the radiation balance of the lower 
stratosphere had assumed that the stratospheric air 
would be at or near saturation. The relative importance 
of water vapour in the radiation balance of the strato- 
sphere is therefore much less than has usually been 
supposed. Dobson [6] has, for example, suggested that 
as a result of the small amount of water vapour pres- 
ent, the carbon dioxide and the ozone, which also take 
part in the radiation processes, may be equally impor- 
tant and that the radiative equilibrium may be deter- 
mined by the relative amounts of these gases present. 
If we consider the water vapour as a “‘tracer,”’ it is 
difficult to give a satisfactory explanation of the pres- 
ence and origin of the extremely dry air of the strato- 
sphere. Since the dry air is consistently found with 
upper winds from all directions, it seems certain that 
the effect occurs in all temperate latitudes, and prob- 
ably in polar regions also. Photolysis of the water 
vapour at these relatively low levels seems unlikely, for 
if it occurred, the reaction would be well known. If the 
photolysis occurs at high levels, it is difficult to see why 
the transition to dry air should be as low as 10 or 12 km. 
The temperature at the equatorial tropopause, which 
is the lowest temperature found in the atmosphere, is 
about 190K and all the results so far obtained might 
be explained by assuming that the air is dried there, 
the excess water being condensed out at the low tem- 
peratures. Cwilong [6] has stated that at very low tem- 
peratures condensation occurs in the form of large ice 
erystals which would fall out easily. 
Air which has been dried to a frost point of about 
190K at the equatorial tropopause can then be brought 
to the polar or temperate stratosphere either (1) by 
ordinary advection parallel to the tropopause, or (2) by 
a zonal circulation in which air rises at the equator, 
enters the stratosphere, and moves to higher latitudes 
THE UPPER ATMOSPHERE 
where it sinks into the troposphere. These processes 
may, of course, occur together in any proportion. What- 
ever the process, it must be considered in conjunction 
with the work of Glueckauf and Paneth [10], who found 
that up to 20 km the proportion of helium in the at- 
mosphere was constant to within 144 per cent (the accu- 
racy of their measurements). This suggests that turbu- 
lent mixing in the stratosphere is enough to maintain 
the composition of stratospheric air constant, at least 
up to 20 km, as otherwise the helium concentration 
would increase rapidly at high levels. The water-vapour 
results, on the other hand, suggest intense stratification. 
Tt seems unlikely that the water-vapour distribution 
is maintained by the advection of dry air from the 
equator without the assistance of large-scale but slow 
zonal overturning, for it should be noted that the flow 
would have to be parallel to the tropopause, and there 
seems no obvious reason for this. Also, in the absence 
of large-scale vertical motions, turbulence is required 
to maintain the constancy of the helium content, and 
on these, and other grounds, the value of K (the diffu- 
sion coefficient) in the lower stratosphere would be ex- 
pected to be at least 10%. In this case, to maintain the 
shape of the observed vapour pressure versus height 
profile, it would be necessary for the air to be redried 
at the equator every twenty or thirty days. Since the 
potential temperature of the air at the equatorial tropo- 
pause is at least 20C higher than at the tropopause 
over England, it would be necessary for the air to be 
cooled by radiation at a rate of one or two degrees 
centigrade per day during its journey from the equa- 
tor; this rate seems improbably high. 
A mean zonal overturning in which air entered the 
stratosphere via the equatorial tropopause in a slow 
but roughly continuous motion would simultaneously 
maintain the dryness of the stratosphere and would 
prevent gravitational settling out of the helium. An 
extremely slow circulation would suffice to account for 
the observed distribution of helium, and the ‘“‘water- 
vapour-height” curves would be maintained against 
the effects of diffusion (K = 10%), with a circulation 
rate such that the mean time taken for the journey 
from the equator is about six months, and with a mean 
subsidence rate in the stratosphere over England of 
about 30 m per day. This would require radiative cool- 
ing of the air in temperate and polar latitudes of 0.3C 
per day. On the other hand, zonal overturning does not 
enjoy current favour and on dynamical grounds it is 
generally believed to be impossible, except in regions 
of zero or negative absolute vorticity. Thus the water- 
content observations raise the whole problem of the 
origin of the stratosphere, first because radiation con- 
ditions are drastically changed by reducing the avail- 
able radiating water vapour, and second because it 
seems unlikely that the dryness of the air can be main- 
tained without significant dynamic movements. If these 
dynamic movements occur, they will affect the tem- 
perature by adiabatic compression or expansion and 
the basic assumption of a stratosphere in radiative 
equilibrium may have to be abandoned. These points 
have been discussed by Brewer [2]. 
