The authors have presented an interesting set of data on the comparison of elastic wave velocities measured on dry and saturated samples of granular soils and some considerations on a longitudinal pulse which has been interpreted as the longitudinal wave of the second kind (Biot-wave). The purpose of this discussion is to make a few notes that seem important in appreciating the signi®cance of the paper.From the comparison of the shear moduli evaluated from the shear wave velocities of dry and saturated samples, the authors conclude that the shear modulus of the saturated samples should be smaller than that of the dry sample and this could possibly be related to the wetting contacts between particles. The authors' analysis has been carried out using quite a low value of the apparent mass density: the`structural factor' á has been assumed equal to 0´25 as recommended by Stoll (1989) from an investigation of the attenuation characteristics of longitudinal waves. However, there are many evaluations of the added mass density, which are based on theoretical analyses, on analogical principles and on experimental measurements (as reviewed by Gajo, 1996), which seem to show that, for granular soils comprised of round grains, the`tortuosity' ô can be approximated bywhere â is the porosity and ô á 1. As a consequence, thè structural factor' á would be approximately equal to 0´06 for coarse Monterey sand, and approximately equal to 0´50 for medium Monterey sand and for ®ne crystal silica sand. These values would reduce the shear wave velocity (of about 2%) and would possibly render unnecessary the reduction of the shear modulus of saturated samples, proposed by the authors in order to have a theoretical shear wave velocity closer to the experimental data. However, these higher values of á would lead to a reduction in the computed velocity of the longitudinal waves of the second kind (about 10%) and this would change the interpretation of all the measurements. It is worth noting, however, that values of á much larger than 0´25 were used by other authors (e.g. Berryman, 1980;Van der Grinten et al., 1987) and led to a good consistency between computed and measured velocities and amplitudes of the longitudinal waves of the ®rst and second kind.Furthermore, the authors have observed the presence of a longitudinal pulse with an amplitude similar to or larger than the amplitude of the longitudinal wave of the ®rst kind (Pwave). This pulse has been considered to be the longitudinal wave of the second kind (Biot-wave). The relatively large amplitude of this pulse is not consistent with theoretical