SUMMARYAnalysis of the results of 104 personnel who determined the particle size distribution of 16 different soil‐size materials indicated that there was an effect of subsample preparation. Subsamples prepared using a chute splitter tended to be coarser than those prepared by coning and channelling. This effect was observed for material retained on 425 μm and 75 μm sieves from a variety of soil types. The results confirm those obtained on artificially prepared mixes. The conclusion is that the chute splitter is a more reliable method for obtaining the test portion than coning and channell