It is suggested that a relationship may exist between the measured sphericity of clasts selected from a sediment and the criterion of size selection employed in the sampling. This relationship is exhibited clearly in cases where the size selection results in a change in the shape of the size distribution of the clasts.
A study of glacial and beach pebbles statistically illustrates the correlation between sphericity and the skewness of the size distribution of the pebbles for each mode of size selection employed.
The results suggest the need for a more critical treatment of sphericity data. It is not unlikely that statistically different sphericities may be largely or even entirely a result of the effect of differing size distributions of the sampled sediments; or conversely, the selection process may augment or obscure any real differences which are present.