TIlE MIXING AND BLENDING OF POWDERS 2ll The mixing of particles of the same size but different densit), Since the mixing of components differing only by colour is very unreal case, the above investigation was repeated using gypsum of specific gravity 2.14 mixed with sand of specific gravity 2.58. The results are presented in Fig. 3 and Table II. F O. I Symbol Run Mixer n R-SVII-$ M-I ß • z• R-SVII-6 M-2 n [] n• • o R-SAVII-I M-2 . • o oo• I I I I I I I I I I 20 40 • 80 I00 120 140 I• I• 200 Mixing time Figure $. Sample Variance S2(t) v. time for sand gypsum ]nixture. Mixer I II II Table I1 Sand: Gypsum. Rotational speed as fraction of critical speed 0.52 0.52 0.33 ]) X 10-2m2sec -t 0.00793 0.00801 0.01440 This table shows results similar to Table I with the exception that at different rotational speed a markedly different value of the migration co- efficient was obtained. It is thus suggested that the scale-up rates suggested

212 JOURNAL OF TttE SOCIETY OF COSMETIC CHEMISTS in the previous section can be used when the specific gravities of the com- ponents are not too different provided that the mixer is rotated at the same fraction of the critical speed. ø,51 Symbol Run o Glass ballofini/sand ß • Zircon/sand Figure 20 40 c•3 80 IO0 12o lOX) 160 180 200 Mixing time Sample variance S2(t) v. ti•nc for sand-zircon and sand-ballotini mixtures. This state, however, must be tested before assuming that different densities will always give this result since when the density ration was increased to 1.81 and sand was attempted to be mixed with zircon sand, strong segregation was found. The heavier zircon particles segregated both radially and axially. A similar state was obtained when glass ballotini was mixed with sand even though here the density ratio was only 1.1, a plot of l n. S2 v t for each mixture is shown in Fig. 4. These results conflict with those of Roseman and Donald (6) who stated that segregation would occur when the density difference was greater than 1.2. Clearly shape is important and has not yet been completely accounted for. Mixing components of different size In order to ascertain the effect of mixing components of different size, two fractions were obtained, S1 the sieve fraction --3•/q-52, and S2 the