Elastic moduli of water-saturated sedimentary rocks are in some cases different from moduli derived using Gassmann fluid substitution on data for rocks in the dry state. To address this discrepancy, we use a data set representing 115 carbonate samples from different depositional settings and a wide range of porosity and permeability. Depositional texture is reflected in the effect of water on elastic moduli and in the porosity-permeability relationship. Depositional texture is taken into account when porosity and permeability are combined in the effective specific surface of pores, which is related for a given pore fluid to the reference frequency as defined by Biot. For a given frequency of elastic waves, we obtain Biot's frequency ratio between measured ultrasonic wave frequency and Biot reference frequency. For mostsamples with a frequency ratio above 10, elastic moduli in the water-saturated case are higher than predicted from elastic moduli in the dry case by Gassmann fluid substitution. This stiffening effect of water in some cases may be described by Biot's high-frequency model, although in heterogeneous samples, a squirt mechanism is more probable. For data representing frequency ratios of 0.01 to 1, Gassmann fluid substitution works well. For samples with frequency ratios below 0.001, elastic moduli in the water-saturated case are lower than would be expected according to Gassmann's equations or to Biot's theory. This water-softening effect becomes stronger with decreasing frequency ratio. Water softening or stiffening of elastic moduli may be addressed by effective-medium modeling. In this study, we used the isoframe model to quantify water softening as a function of frequency ratio.