The number median diameter (NMD), geometric standard deviation (o,), and average diameter (Davg) wére obtained for each sample anilyzed. The NMD is defined as that size such that 50 per cent of the number of the particles are smaller and 50 per cent are larger than the stated size. The value is obtained by interpolaticn of two vals bracketing the 50 per cent line on a cumulative graph of number distribution. The geometric standard deviation (¢ ) is a masure of the de- gree of homogeneity of the sample. following relationships .21/ It is defined by either of the : cumulative 84.14 percent particle size on log-probability plot Og = cumilative 50 percent particle size on log-probability plot (3.24) °g= eumlative 50 percent particle size onlog-probability plot cumulative 15.87percent particle size on ee Os plot 025 The range from 15.87 percent to 84.13 percent is cne standerd deviation. %g may theoretically be any valve from 1 to infinity. Values neer 1 indicate a homogeneous sample, As the value increases, samples are indicated as being more hetrogeneous. In practice, values rarely are higher than 4 to 6 for field samples. The average diameter Day, = 2Dn Sn (3.26) where 2D, is the sum cf the diameter of all of the particles 2nis the sum of the number of particles Particles as large as 3000 m were found during the analysis. The procedure of separating each sample into two fractions eliminated the requirement of a common exposure time for both small and large particles and the smaller particles were mre easily distinguished than they would be in an unfractionated sample. Since a gross particle size distribution was not made, the data from both fractions of each sample could not be recombined to give one NMD for each station. However, the number of particles in the larger fraction was found by microscopic examination to be only a smll percentage of the number of particles in the gross sample; hence, the small fraction NMD would not be raised by any great extent, if it had been possible to combine the two fractions. Thus, the NMD of the small fraction may be considered to be the approximte NMD of the entire sample. It should be pointed out that the use of sieves in fractionating particles may have some tendency to break up agglomerated particles into their smiller com ponents, although some experimental evidence indicates that this effect is minor. Particle size results are presented in Tables 3.12 and 3.13 and are summarized as follows: SHOT 1: The NMD of the small fraction ranged from 5 to 17.5. The NMD of the lerge fraction ranged fro: 61 to 8p. 86