THE PHYSICS OF ICE CLOUDS AND MIXED CLOUDS 
small ice-supersaturations). Fournier d’Albe [9] pro- 
duced these particles by evaporating crystals which 
had been grown at —41C im the presence of solution 
droplets of cadmium iodide. It seems likely that the 
particles remaining consisted of cadmium iodide bearing 
microfilms of water which retained an icelike structure 
such as cannot be imposed upon films adsorbed from the 
vapour, even by cooling to temperatures in the neigh- 
bourhood of —70C. This suggests that the only sublima- 
tion nuclei are in fact particles of ice, and raises the 
possibility that after the evaporation of ice clouds at 
low temperatures some freezing nuclei may be left in a 
condition which enables them to act as sublimation 
nuclei in a future cloud formation. However, nuclei 
having the efficiency of those made by Fournier d’Albe 
cannot be expected to occur naturally in the atmos- 
phere, where even the poor ice nuclei available are 
greatly outnumbered by nuclei capable of aiding the 
formation of liquid droplets. This scarcity of ice nuclei, 
which may be even more marked in the high troposphere 
than near the ground, is an important factor which in 
association with the speed of atmospheric condensation 
processes results in the frequent appearance of droplets 
during cloud formation at temperatures down to —50C 
or even lower. If the humidity is creased very slowly 
some ice particles will form some time before water- 
saturation is reached; their subsequent growth may then 
remove enough vapour to prevent a further increase in 
humidity, or the process responsible for the rise of 
humidity may cease. In this way very thin pure ice 
clouds may arise without any droplet formation. Their 
occurrence must be very infrequent in view of the great 
scarcity of ice nuclei which act before water-saturation 
is almost reached, and the slowness and duration of the 
condensation mechanism which would be necessary, but 
the structureless veils of cirro-nebula and the ‘‘diamond- 
dust” ice-mists probably are such clouds. 
Evidently, in experiments designed to discover the 
critical temperatures at which droplet formation is 
entirely replaced by crystal formation in the presence 
of foreign nuclei, particular attention must be paid to 
the speed of the condensation process. It is clear that 
the violent expansions commonly used in cloud cham- 
bers may produce predominantly droplet clouds at 
temperatures down to —40C and even lower, and it is 
also clear why different workers, using various rates of 
expansion, have found different values for the critical 
temperatures. 
The expansion-chamber experiments which most 
faithfully reproduced natural conditions are those of 
Findeisen and Schulz [8], who used an unusually large 
chamber (volume 2 m*) to increase the sensitivity to 
the rare but still meteorologically important ice nuclei, 
and who controlled the rate of expansion to values 
corresponding to vigorous atmospheric convection 
processes. Their apparatus was not portable, and so only 
the surface air (at Prague) was examined for ice nuclei. 
The number of such nuclei in this polluted air may 
have been much larger than is usual in the air of the 
upper troposphere, but even so in the chamber-clouds 
the droplets still outnumbered the crystals at temper- 
193 
atures around —40C. Down to this temperature the 
clouds formed in the chamber by uninterrupted ex- 
pansions were initially pure water-clouds or mixed 
clouds, but some pure ice-clouds were produced with 
great difficulty by ceasing the expansion just before 
droplet-formation was expected. The numbers of nuclei 
active at various temperatures composed an ice-nucleus 
“spectrum,” which changed somewhat from day to day 
and with the rate of expansion, but which possessed 
the general form shown in Fig. 1. With another ap- 
NUMBER OF EFFECTIVE NUCLEI cm? 
-20 =15 -10 =I) te) 
TEMPERATURE °G 
Fic. 1—Average ice-nucleus “spectrum” for samples of sur- 
face air at Prague, showing the numbers of nuclei active at 
various temperatures during the expansion of the air at rates 
corresponding to vertical speeds in the atmosphere of 5 msec? 
and 20 m sec. (After Findeisen and Schulz.) 
=35 
paratus, in which the speed of the condensation process 
was much less, Schulz [17] has recorded results which 
do not differ materially from those given in this figure. 
As the temperature falls below about —32C there is a 
sudden great merease in the number of active nuclei, 
which Findeisen and Schulz consider due to the intro- 
duction of a second class of nuclei. These appear to be 
the nuclei responsible for the first appearance of crystals 
(in droplet fogs) in the smaller chamber used by 
Fournier d’Albe [9], and which are now referred to as 
the 32-nuclez. Findeisen and Schulz apparently did not 
continue their experiments to temperatures below 
—40C, and so did not find a second critical temperature 
of about —41C below which Fournier d’Albe observed 
relatively dense ice-fogs in which the presence of any 
droplets could not certainly be established. Fournier 
d’Albe, however, was able to confirm that these ice 
fogs formed only when the air was cooled near to, if not 
below, the dew point (at saturation with respect to 
liquid water). The reason for this critical temperature 
of —41C and the part it plays in more moderate ex- 
pansions remain unknown. More detailed examination 
may show this critical temperature to be no more sharply 
defined than the threshold temperature at which the 
32-nuclei become active. 
Very little is known about the number of ice nuclei 
occurring in the upper troposphere. Palmer [14] has 
