
Supplement to “ Nature,” April 14, 1923 
+e 
vil 

supersaturation. A single molecule of a hygroscopic 
substance would probably be able to gather around 
itself sufficient water to make a drop large enough to 
grow by ordinary condensation. Thus, whereas the 
hygroscopic properties are important to build up a drop 
to a certain size, after that size has once been reached 
the hygroscopic attraction may cease through excessive 
dilution, but condensation will continue until the drop 
is in equilibrium with the surrounding vapour. 
The ordinary small ion, in my opinion, takes no 
part whatever in meteorological processes. Before any 
deposition of water can take place on small ions, four- 
fold supersaturation is necessary in the case of negative 
ions and six-fold in the case of positive ions. We have 
absolutely no evidence of anything like these super- 
saturations in the atmosphere, and I have shown at some 
length, in a paper published in the Phtlosophical 
Magazine, that even in the case of thunder-storms, in 
which the conditions are by far the most favourable 
for the formation of supersaturated air, small ions take 
no part in the condensation. Further, we have no 
evidence that small ions act like hygroscopic substances; 
they do not appear to grow appreciably in size in 
saturated air, in fact they act like any other air molecule 
until four-fold or six-fold supersaturation is actually 
reached. 
With regard to large ions (Langevin ions), these do 
appear from Pollock’s work to grow with increase of 
humidity ; but, as Pollock found that they do not form 
in pure air, these ions are probably nothing more than 
ordinary hygroscopic nuclei, with a small ion attached. 
Without detracting in any way from the value of 
Aitken’s work, we see that it is necessary to revise his 
conclusions, and to say that hygroscopic substances 
and not dust form the nuclei of condensation. I do not 
think that Aitken would have been surprised at this 
development of his work, for he clearly recognised the 
importance of salt particles as efficient nuclei. 
CONDENSATION AT TEMPERATURES BELOW THE 
FREEZING Point. 
When the temperature of the air is below the freezing 
point, condensation of the contained water vapour is 
a still more difficult problem, for there is very little 
experimental evidence to go upon. One thing is certain: 
owing to the small amount of vapour present, it is 
inconceivable that condensation will take place by the 
fortuitous meeting of molecules; some kind of nuclei 
therefore will be necessary. 
According to experiments made on crystallisation 
from supersaturated solutions, we may conclude that 
_crystallisation does not start readily on a perfectly 
spherical surface. An angular nucleus is necessary, and 
the nearer the angles are to those of the natural angles 
of the crystal the more readily will condensation take 
place. When sledging in the Antarctic with Captain 
Scott in 1911 we became enveloped in a fog during 
sunshine. On the fog opposite the sun we saw a white 
bow in the position usually occupied by a rainbow. 
This phenomenon can only be explained on the assump- 
tion that the fog was composed of small spheres ; but 
the temperature was — 29° C. (— 21° F.). We are quite 
familiar with super-cooled water drops which have been 
formed when liquid drops are cooled from temperatures 
above to temperatures below the freezing point. In 
this case there was no part of the atmosphere within 
hundreds of miles of the place of observation in which 
the temperature was above the freezing point, and 
almost a dead calm existed at the time; hence these 
drops could not have formed at a high temperature and 
then been super-cooled. 
The only explanation which appears possible to 
me is that in the clear air of the Antarctic there 
are no “dust” particles suitable for condensation 
available, but there are plenty of the hygroscopic 
molecules of which we have already spoken. With 
increasing humidity these “hygroscopic molecules 
attract water from the air and form clusters of water 
molecules ; but we know from colloidal chemistry that 
such clusters are in the spheroidal and not in the 
crystalline form. If this explanation is correct 
we have real liquid drops and the vapour pressure 
in the air must be that appropriate to a water 
surface—we may neglect the curvature of the drops, as 
their radii were probably of the order of o-oor cm. But 
according to Table II. air at — 30° C. is saturated with 
reference to ice at a relative humidity of 74 per cent., 
hence in this case the air was heavily supersaturated 
with reference to the solid state. This accounts for 
the fact, recorded at the time, that “the fur of the 
sleeping-bags and the wool of sweaters became covered 
with hoar frost,” for these substances forged excellent 
nuclei for the condensation of the water vapour direct 
into the solid state. 
Support for this explanation is given by observa- 
tions made by Wegener in Greenland. He describes 
how, in temperatures below — 40° C. with perfectly 
clear air, a strip of fog started at the house and extended 
for several kilometres in the direction of the wind. At 
such temperatures condensation takes place on water at 
roo per cent. relative humidity, but on ice or solid nuclei 
at 67 percent. The actual humidity was between these 
limits, therefore the air was not saturated with reference 
to the hygroscopic nuclei, but was supersaturated with 
reference to solid nuclei. There were, however, none of 
the latter present in the pure free air, but the air 
escaping from the hut was highly charged with solid 
nuclei, chiefly the products of combustion, and when 
