ARTIFICIAL CLOUD-MODIFICATION AND PRECIPITATION 
optimum ratio because of the variability of the water- 
droplet distribution in different stratus cloud systems 
and because of differing diffusion conditions. It is obvi- 
ous, considering the tremendous number of nuclei cre- 
ated from the fall of one dry-ice pellet, that there will 
be gross overseeding in the cloud segment through 
which such a pellet falls while at some distance away, 
depending upon diffusion conditions, there will be un- 
derseeding. Somewhere, then, between where the pellet 
passes through the cloud segment and where its nu- 
cleation effects are just felt, there must exist the opti- 
mum ratio for the particular cloud. It is seen, then, that 
the problem of overseeding is not a consequence of ex- 
cessive amounts of dry ice thrown out along a path, but 
rather is due to the lack of uniform dispersion by the 
diffusion of the tremendous numbers of nuclei pro- 
duced by even one dry-ice pellet. To quote Langmuir 
[12] “The conclusion is obvious, however, that with a 
reasonable number of pellets dropped along a flight 
path into the top of a cloud, the limiting factor will 
not be the number of nuclei but the rate at which the 
nuclei can be distributed throughout the cloud.” In 
cumulus clouds, the characteristic internal turbulence 
and mixing would spread the nuclei from even one 
pellet throughout a large portion of the cloud within a 
short period of time. However, in characteristically 
stable stratus clouds, only horizontal divergence or 
mixing would tend to distribute the nuclei through a 
large cloud volume. These factors are of course the 
same ones that result in natural dissipation of stratus 
clouds. Thus, the results of the Cloud Physics Project 
emphasize the importance of divergence and mixing 
by diffusion. In both the Ohio and California tests, 
spread of artificially imduced nuclei and subsequent 
stratus dissipation were observed only when natural 
dissipation was occurring. The effect of heats of fusion 
and sublimation released as a result of seeding was 
never observed to be on a scale large enough to cause 
any important local circulation and resulting spread 
of artificial nucleation. 
Experiments with Cumulus Clouds 
It appears that the most suitable cloud for the arti- 
ficial production of precipitation would be a cumulus 
congestus whose top, having reached above the freez- 
ing level, would contain supercooled water while still 
not containing ice crystals. If seeding resulted only in 
the conversion of the water in such a cloud into precipi- 
tation, it is possible that a maximum of 0.2 in. of rain- 
fall might reach the ground under the most favorable 
circumstances. In addition, it has been suggested that 
the heat released in the conversion of the cloud top to 
ice crystals might be of sufficient value to set off addi- 
tional convection within the treated cloud and thus to 
result eventually in the production of a heavy rain 
shower or thunderstorm. 
The investigation of the results of seeding cumulus 
clouds is somewhat more difficult than the investiga- 
tion of similar results in stratus clouds. In the latter, 
the untreated portion of the seeded deck can be used 
as a control or a standard, while the ever-changing 
239 
configuration and often violent activity of cumuli make 
it impossible to use any one of them as a control. It 
has been suggested that the cloud to be seeded be 
chosen randomly from among those possessing the re- 
quired characteristics and that, for each cloud seeded, 
a similar one, also chosen randomly, be observed in 
its natural course. However, the results of the Cloud 
Physics Project in seeding cumulus clouds indicate that 
the control exercised through the use of the several 
observational aircraft and, especially, of a rain-sensi- 
tive radar, made the scheme described above unneces- 
sary. Altogether the project conducted 79 seedings in 
cumulus clouds in the Ohio area [5] and 44 in semi- 
tropical cumulus along the Gulf Coast [7]. A number 
of clouds seeded in the Ohio tests were not supercooled 
but were seeded with water droplets for the purpose of 
investigating the reality of “chain reaction processes” 
within warm clouds as envisaged by Langmuir in 1948 
[10]. The results of seedings in such warm clouds ap- 
peared to be unfavorable from the relatively few tests 
conducted, and they were not continued in the Gulf 
Coast tests, but rather the more favorable supercooled 
cumulus clouds were investigated there. 
As indicated in U. 8. Weather Bureau Research 
Paper No. 31 [5], only trivial amounts of precipitation 
resulted from seeding the Ohio cumulus clouds and 
even in cases where these amounts were observed, 
there were nearby natural showers of the same or 
greater intensity. In none of the cases reported did the 
seeded cloud build extensively as a result of the dry- 
ice drop and it is thus obvious that no large amounts 
of precipitation could have been expected to fall. On 
the contrary, the most obvious effect of seeding dry 
ice into both supercooled and nonsupercooled cumulus 
clouds was the resulting rapid dissipation. There were 
very few cases in which the cloud built up after seed- 
ing, and in each of these the growth did not exceed a 
few thousand feet. On the other hand, dissipation was 
usual and often almost complete in its effect. The fol- 
lowing is quoted from this report [5]: 
This dissipation appeared to be nearly independent of super- 
cooling or of the particular agent employed. Its occurrence 
was consistent with the idea that convective clouds often 
have lapse rates steeper than the moist adiabatic as a result 
of mixing and entrainment between such clouds and the en- 
vironment. A downward movement initiated by dry ice, large 
numbers of ice crystals, water or other means (aircraft flying 
vertically upward through the cloud have been employed 
successfully) might easily cause an appreciable mass of air to 
become colder than the surrounding clouds, and thus induce 
further downward motion. A similar explanation has been 
advanced by Byers and Braham for the formation of the 
thunderstorm downdraft. 
Regarding the general results in the Ohio area, the 
following is also quoted from the same report: 
The experiments showed that the artificial modification of 
cumuliform clouds is of doubtful economic importance for 
the production of rain. Dissipation rather than new develop- 
ment was the general rule. There is no indication that seeding 
will initiate self-propagating storms, and therefore, the only 
precipitation that can be extracted from a cloud is that con- 
