In papers I and II of this series, it was concluded on the basis of theory and a limited number of experimentalsamples that the over-all differential probabilityp′of single fracture, as a function of the mean fragmentdimensionx, should show the presence of three peaks. These correspond to the effect of internal flawsof facial and volume type, and of surficial flaws. This result is verified experimentally with good statisticsfrom the mean of data for 36 specimens fractured in gelatin by Bergstrom, Sollenberger, and Mitchell. Theeffect in this over-all distribution of the exoclastic fragments produced by surficial flaws was subtracted out.The resulting endoclastic differential probabilitypfor the 36 specimens, corresponding to the effect of internalflaws only, shows with good statistics the presence of two peaks arising from facial and volume flaws.The result of I that the number of fragments cumulated over fragment volumes down to a lower limitvshould vary asv−23for small fragment size is confirmed by measurements for specimens that underwentsingle rather than plural fracture.