of particle agglomeration during dispersal 
owing to the serious effects in particle size 
distribution determination by automatic scan- 
ning. 
In optical photomicrography for automatic 
scanning, the large amount of light usually 
transmitted by fine particles andthe inherently 
short focal depth present considerable prob- 
lems, Hence, the technique originated by 
Hamilton and Phelps (13) is utilized, After the 
particles are Suitably dispersed on a mciro- 
scope slide, a negative image of the dust 
deposit is prepared by vacuum evaporation of 
metal at normal incidence onto the slide, When 
the slide has received a precise, natural-size 
impression of the geometric projections of 
the particles, the particles are removed, 
Transparent profiles or holes in the metal 
film remain as ''shadows"' cast by the particles, 
The technique is applicable only to samples 
where the particles are deposited directly on 
the slide without the use of an adhesive 
material, Since the metal film is at most only 
a few hundred angstroms thick, a sharp plane 
of focus is provided for the optical micro- 
scope, Adjustment of film thickness aids at- 
tainment of the desired degree of contrast 
in the ultimate photomicrographs, Copper has 
been found to be a very satisfactory shadowing 
metal, since it exhibits no objectionable struc- 
ture effects in optical photomicrography, Cer- 
tain particulate substances aresometimes dif- 
ficult to remove subsequent to the shadowing 
process, and work is underway to deal with this 
problem, Among the removal methods being 
investigated are the application of air and 
water jets, as used by Hamilton and Phelps, 
the use of wetting agents, and ultrasonic treat- 
ment, 
Either transmitted or incident microscope 
illumination may be used for the metal- 
shadowed specimens where, under visual ob- 
servation, particles appear as bright areas in 
a dark field or as dark areas ona bright field, 
respectively, Although the particle analyzeris 
capable of measurements of either dark par- 
ticles on a light field or the reverse, ultimate 
selection of the illumination method rests upon 
the quality of photomicrographs obtained, 
Attempts are being made to use automatic 
scanning on electron micrographs of clay par- 
ticles, Particles are dispersed on a specimen 
screen and shadowed at relatively shallow 
205 
incidence by means of chromium evaporation; 
the particles are left in place subsequent to 
shadowing, After preparation of micrographs 
in an electron microscope, photographic pro- 
cedures are used to further enhance contrast 
and to reduce images to dimensions compatible 
with film requirements of the particle analyZer, 
Although some degree of success has been ob- 
tained, much further work is indicated before 
the methods can be considered suitable, 
Diffusion of Particles in Turbulence 
At the U.S, Department of Agriculture Pio- 
neering Research Laboratory on Physics of 
Fine Particles, theoretical and experimental 
studies are in progress on the behavior offine 
particle suspensions in fluids, The basis and 
elements of the approach used are discussed 
by Pasquill (21) and Sutton (24), and are briefly 
outlined here, 
Most of the early theoretical treatments of 
turbulent diffusion have been based on the 
phenomenological mixing-length or transfer 
theory, as developed by Prandtl and others, 
In such theory, a partial differential equation, 
similar to the ordinary diffusion equation, is 
used to describe the particle dispersion 
process, The counterpart of the usual dif- 
fusion coefficient is a parameter termed the 
eddy diffusivity, or exchange coefficient, A 
large difficulty, particularly in the case of 
atmospheric diffusion, has been to adequately 
describe the variation to the eddy diffusivity, 
which oc¢urs with respect to position (e.g., 
elevation) in the atmosphere. 
However, a factor of more critical im- 
portance is that modern investigations in 
turbulence and turbulent diffusion have made 
clear more serious shortcomings inthetrans- 
fer theory, For example, the transfer theory 
of turbulent diffusion assumes that a gradient- 
type diffusion prevails in the turbulent fluid, 
which does not necessarily follow in the light 
of statistical theory, Also, diffusion and con- 
vection of turbulent energy are not negligible, 
as the phenomenological theories presume, On 
the other hand, although the phenomenological 
theories are not completely satisfactory from 
a more modern research point of view, they 
are at present the most useful in practical 
applications, Since the statistical theories have 
