In a series of physical model experiments, fractured media are simulated by stacks of thin Plexiglas sheets clamped together tightly to form blocks. The plates are assembled underwater, and a very thin water layer between the sheets prevents formation of an effectively welded interface between them. Thus, the stacked material is not a series of welded plates but rather a truly fractured medium simulating a potential petroleum reservoir with only fracture porosity and permeability. Sheets of constant thickness are used, but the intensity of fracturing between the different models is simulated by using different thicknesses of Plexiglas for each model. Observation of direct shear-wave arrivals through the stack, with propagation parallel to the sheets and polarization of particle motion allowed to be parallel to, normal to, or in any arbitrary angle to the sheets, definitively demonstrate the existence of shear-wave splitting and hence anisotropy. For Plexiglas sheets 1/16 in thick (0.16 cm), representing a fracture intensity of about 16 fractures per wavelength, shear-wave splitting and hence anisotropy are clearly observed. For greater fracture intensities (i.e., thinner plates), the degree of anisotropy is greater, and for less intense fracturing (i.e., thicker plates), the degree of anisotropy is less. These experimental data suggest that for fracture intensities of greater than about 10 fractures per wavelength there is a simple relation between fracture intensity and degree of anisotropy in two different polarizations of shear waves. For fracture intensities less than a threshold of about 10 per wavelength, there is no simple observed relation between them. Further, the faster of the split shear-wave velocities is near the solid material velocity, and the observed variation in velocity for various fracture intensities is in the slower of the split shear-wave velocities.