samples that have been size-separated in the laboratory. Mass or volume distributions can be derived from the size distributions but have primarily been determined by the weighing of size fractions obtained by sampling in the field with cascade impactors, or in the laboratory by sieve analysis. For impactor sampling, the applicable size range is determined by the first stage collecting all particles larger than about 10 ym and by the backup filter collecting the particles that are smaller than about 0.5 ym, Sieve analysis is useful for samples consisting mostly of large particles, the smallest sieve opening used being 44 um. Samples, or sieve fractions, consisting mostly of small (<50 um) particles may be separated into size fractions by allowing the particles to settle in an appropriate liquid having a small density gradient to eliminate convection currents. The particles that collect at the bottom of the liquid column are removed at predetermined times. |000 500 As many as 14 size Particle size measurements are always made on gmall aliquots of the samples, usually on filter collections, Sections of the filters were ashed, the ash was dispersed in an appropriate liquid. Microscope siides or electron microscope grids were prepared from the dispersion. Individual particles were measured in the light microscope by means of a calibrated scate, later also by means of an image-splitting device or a Porton graticle. Whereas in the first two methods, the actual sizes of individual particles were measured, the use of the graticle places the particles in size bins and effectively determines a histogram with logarithmicly equal size intervals. For irregular particles, the definition of size is not always clear, and may differ among investigators. The geometric mean of the Largest and smallest diameter has been used to define size. Size has also been defined as the length of the projection of the particle on an (imaginary) horizontal line, Thickness has never been considered. DIAMETER (,L) fractions have been thus obtained (Heft and Steele, 1968). High-speed centrifugation has also been used. The size distribution of the fractions are added with the fraction weights as weighting factors. These fraction weights are obtained by filtering the fractions and weighing, or by chemical analysis. Figure 1 shows the size distributions of one such set of size fractions. The overlap of the size ranges may have been caused by such factors as shape and density variations, entrainment of small particles by large particles, and so forth. Clearly, none of these size and mass distribution measurements distinguish between radioactive and nonradioactive particles. In a few cases, e.g., with balloon tests and one test buried at a shallow depth, a distinction was made between fused smeoth spherical or nearly spherical particles and irregular particles to obtain separate size distributions (Russell, 1965; Nathans et al., 1970). The spherical particles were assumed to be primary debris particles. They are significance only below about 5 ym. On a few samples from other tests, radioactive and nonradioactive particles were distinguished by autoradiography, and size distributions of these radioactive particles were measured separately. This procedure is, of course, extremely cumbersome. 1 1 Figure 1. 232 1 2 Po. § 10 bd dd 20 3640506070 8 90 95 9899 CUMULATIVE PERCENT Particle size distribution of size fractions from a Johnie Boy sample 233