SNOW AND EXPERIMENTAL METEOROLOGY 
the Great Plains which were carried aloft as the air 
passed over those regions. 
The very important conclusion which may be in- 
ferred from the Mount Washington studies is that for 
extended periods the atmosphere contains very low 
concentrations of suitable nuclei for the formation of 
ice crystals. Thunderstorms, hail, heavy rain, and other 
storms which have part of their structure above the 
OC isotherm are basically related in their developmental 
stage to the concentration of ice nuclei in the atmos- 
phere where they form. It follows, therefore, that any 
modification in the natural concentration of such par- 
ticles m the atmosphere should exert profound effects 
on the formation of such storms. This subject will be 
discussed in greater detail later under the section on 
experimental meteorology. 
The most effective sublimation nucleus known at the 
present time is silver iodide. This fact was discovered 
during research activity in the General Electric Re- 
search Laboratory [37]. Since silver iodide does not 
normally occur in the free atmosphere, and since it is 
much more active than any known natural ice nucleus, 
far-reaching and anomalous effects may be expected to 
occur when large numbers of these particles are released 
in regions where supercooled clouds occur. 
As Vonnegut [89] has shown, silver iodide is a very 
effective nucleus for ice formation because its atomic 
and crystalline characteristics, including the size of the 
unit cell, approach the structure of ice within 1 per 
cent. 
While the temperature dependence of the effective- 
ness of sublimation nuclei is probably related to a num- 
ber of such parameters, the nature of the surface layers 
of molecules seems to be of primary importance in de- 
termining whether a surface will be kryophilic or kryo- 
phobic. 
Typical effects illustrative of these properties are 
shown in Fig. 7. The kryophilic surface (Fig. 7a) con- 
sisting of clean glass, is coated with frost crystals whose 
Major, or c-axis, is perpendicular to the glass surface. 
The kryophobic surface (Fig. 7b) is glass treated with 
a molecular layer of a dichlorosilane. Over the relatively 
small area contacted by frost crystals, the c-axis is 
parallel or inclined away from the treated surface at 
an acute angle. This surface is so strongly kryophobic 
that less than 1 per cent of the glass is contacted by the 
frost crystals. 
It is my belief that such properties are inherently 
related to the effectiveness of, for example, a clay par- 
ticle (active as an ice nucleus at —15C) and the spore 
of Lycoperdon gemmatum (which does not act as a 
crystal center until the temperature becomes —36C). 
3. Spontaneous Nuclei. At high levels in the atmos- 
phere where cirrus clouds form, or at lower levels when 
the temperature is below —39C, foreign particles are 
not required for the formation of ice crystals. When- 
ever moist air supersaturated with respect to ice is 
cooled below —39C tremendous numbers of ice crystals 
2. These terms are suggested to designate the tendency of a 
surface to permit or to prevent, respectively, the formation 
of a frost layer. 
225 
appear spontaneously. One of the easiest ways of dem- 
onstrating this effect is to use the cold-chamber method 
previously mentioned [29]. Tiny dry-ice fragments 
rats V7 
Fig. 7.—Frost erystal structure produced by surface prop- 
erty of a solid surface: (a) crystal forms on clean glass, and 
(6) crystal forms on glass coated with a very thin layer of a 
dichlorosilane. 
dropped into cold air supersaturated with respect to 
ice produce many millions of ice crystals, as shown in 
Fig. 8. Photomicrographs of such crystals are shown in 
Fig. 8.—Snow crystals forming in a cold chamber with air 
supersaturated with respect to ice. 
Fig. 9. It can easily be shown that 1 g of dry ice may 
generate 10!* ice crystals under such conditions [26]. 
The critical temperature which produces this effect 
may be determined in a simple manner. By solidifying 
