194 
used an expansion chamber in an aircraft on four 
occasions and found that the 32-nuclei were absent (or 
very scarce) above the lowest inversion or haze-top. 
This suggests that these nuclei at least are derived 
from the earth’s surface, and that at the cirrus levels 
the ice-nuclei count is likely to be far smaller than that 
observed at the surface by Findeisen and Schulz [8]. 
It is therefore to be expected that, in the atmosphere, 
clouds formed by the more vigorous ascending motions 
will initially contain liquid water even at very low 
temperatures, and indeed this has been observed in 
several ways. For example, R.A.F. aircraft have re- 
ported serious icing at temperatures down to below 
—40C [13], Weickmann [24] has photographed droplets 
(with crystals) at —50C over the Alps in the lenticular 
Moazagotl cloud, and aircraft wing-icing has been ob- 
served at —58C in a condensation trail formed at the 
airscrew [1]. The limit of supercooling, at which crystals 
grow spontaneously in liquid water without the aid of 
foreign nuclei, is not known. It has been roughly esti- 
mated theoretically at about —70C or below, and Rau 
[15] appeared to have found it experimentally at —72C, 
but Cwilong [4] has since thrown doubt on this result. 
It would certainly seem that at all temperatures likely 
to occur in the troposphere ice crystals always require 
special nuclei for their formation, and that these nuclei 
are always actively available at low enough temper- 
atures, since stable supercooled clouds have apparently 
not been observed at temperatures below about —385C 
[24] and are rarely observed at temperatures below 
about —15C. 
Before leaving the discussion of nuclei it may be said 
that further information is needed on the behaviour of 
hygroscopic nuclei at low temperatures, both when they 
are pure and when they are contaminated by freezing 
nuclei. Wall [20] has considered the matter theoretically, 
and some experiments on the freezing of droplets of 
common-salt solution are described by Weickmann [24]. 
From these latter experimentsit appears that salt crystals 
can act as ice nuclei at temperatures below about —30C, 
but im spite of the decreasing solubility of salt with 
fallimg temperature it is unlikely that solid salt crystals 
can occur at high humidities unless they arise in con- 
taminated solution-nuclei. It also seems that freezing 
nuclei may have their efficiency considerably reduced in 
the presence of a concentrated salt solution, so that 
such nuclei might begin to act only when the solution 
becomes sufficiently diluted by condensation. Mason, at 
the Imperial College of Science and Technology Gn 
work unpublished), has attributed the critical temper- 
ature of —41C mentioned above to the freezing of 
solution droplets in which the ions formed by the disso- 
ciation of the electrolyte act as nuclei. 
Meteorologically, the most important features of the 
recent studies on ice nuclei are the recognition of the 
rarity of ice nuclei and the establishment of the fact 
that ordinarily they allow the formation of ice crystals 
only near water-saturation, and therefore in the 
presence of high ice-supersaturations. Present problems 
concerning ice nuclei may be divided into two classes. 
1. Problems related to individual nuclei: What is the 
CLOUD PHYSICS 
mechanism of nucleation? and, What are the properties 
which determine the efficiency of an ice nucleus? Such 
questions may be answered by examining the behaviour 
of small numbers of artificial nuclei, or perhaps even a 
single nucleus, under humidities increased at carefully 
controlled rates. Dew-plate techniques are likely to be 
. more convenient than those using expansion chambers, 
in which the control of temperature and humidity is 
very difficult, but investigations must first be made to 
determine whether the presence of a supporting surface 
interferes with the behaviour of the nuclei. If it can be 
established that certain unwettable surfaces are suffi- 
ciently indifferent, this method could readily be used to 
examine the efficiency of particular nuclei when they 
are covered by films of adsorbed vapour, and when they 
are immersed in droplets of both pure water and salt 
solutions. It should be possible to determine whether 
solutions of salts or of gases (e.g., NO, SOs, and NH3) 
are in any circumstances able to act as ice nuclei when 
not contaminated with insoluble solid particles. 
2. Problems of a more meteorological nature: What 
are the effective ice nuclei in the atmosphere? and, 
What are their numbers at different levels in various 
air masses? To observe the numbers of ice nuclei oc- 
curring naturally a very sensitive counter is required, 
capable of indicating ice-nucleus concentrations as low 
as 1 m-*. The high-pressure expansion apparatus de- 
seribed by Schulz [17] was designed (but never used) as 
a portable nucleus counter; in this apparatus the air 
was first compressed to a pressure of about twelve 
atmospheres so that, on controlled expansion to about 
one atmosphere, low temperatures could be reached 
without the necessity of a preliminary cooling of the 
expansion chamber, while the concentration of ice nuclei 
was not reduced below that occurring in the atmosphere. 
However, elaborate precautions are needed to prevent 
the introduction of artificial nuclei during the compres- 
sion process, and it is to be hoped that some much 
simpler counting apparatus may be practicable. 
THE DEVELOPMENT OF THE ICE PHASE IN 
NATURAL CLOUDS 
Cumulus and Cumulonimbus. Cumulus clouds are 
composed of air which has risen from near the ground, 
and conditions within them are likely to resemble 
closely those produced in the expansion chamber of 
Findeisen and Schulz [8]. Accordingly, ice-crystal forma- 
tion does not begin in these clouds until the temperature 
has fallen several degrees below 0C, depending to some 
extent on the rate of ascent of air mto the cloud base 
(see Fig. 1). Since the crystals are in a considerable 
supersaturation they grow more rapidly than the drop- 
lets and after a period of perhaps a few minutes reach 
a size and falling-speed such that growth by coagulation 
with droplets in their path begins to exceed that due to 
the condensation of vapour. Rough calculations show 
that this stage is reached when the ice particles have 
reached diameters of about 100 uw and fallmg-speeds of 
about 14 m sec~! and when (relative to the surrounding 
air) they will have fallen rather less than 30 m. Beyond 
this stage growth by coagulation rapidly accelerates, 
