Abstract The experimental results and the widely used empirical formula (E/EApp)=(1+βS2)−1+(E/B) for the compression modulus are compared with analytical solutions obtained by Moghe and Neff. It is shown that the empirical formula does not represent the data adequately, particularly for large and small shape factors. It is also shown that the variation in β in the empirical formula arises due to the inadequacy of the representation rather than the thixotropic behavior as claimed.